@SuppressWarnings("static-access")
@Override
public EventSet createEventSet(Document doc) {
EventSet es = new EventSet(doc.getAuthor());
char[] text = doc.getProcessedText();
String stringText = new String(text);
// use MaxentPOSTagsEventDriver's tagger
// initialize tagger and return empty event set if encountered a problem
if (tagger == null) {
tagger = MaxentPOSTagsEventDriver.initTagger();
if (tagger == null) return es;
}
List<List<HasWord>> sentences =
tagger.tokenizeText(new BufferedReader(new StringReader(stringText)));
ArrayList<TaggedWord> tagged = new ArrayList<TaggedWord>();
for (List<HasWord> sentence : sentences) tagged.addAll(tagger.tagSentence(sentence));
int i, j, n;
try {
n = Integer.parseInt(getParameter("N"));
} catch (NumberFormatException e) {
n = 2;
}
String curr;
for (i = 0; i < tagged.size() - n + 1; i++) {
curr = "(" + tagged.get(i).tag() + ")";
for (j = 1; j < n; j++) {
curr += "-(" + tagged.get(i + j).tag() + ")";
}
es.addEvent(new Event(curr));
}
sentences.clear();
sentences = null;
return es;
}
