private void doParseSentencesInCorpus(File ntvmsnbcCorpus) throws IOException {
/* SentenceMorphParse parse = parser.parse("Turgut Özal'ın ölüm raporu ile ilgili flaş bir gelişme.");
parse.dump();
System.out.println("After disambiguation:");
parser.disambiguate(parse);
parse.dump();
for (SentenceMorphParse.Entry entry : parse) {
System.out.println(entry.input + "=" + entry.parses.get(0));
}
for (SentenceMorphParse.Entry entry : parse) {
System.out.println(entry.input + " kök=" + entry.parses.get(0).stem);
}*/
List<String> sentences = SimpleTextReader.trimmingUTF8Reader(ntvmsnbcCorpus).asStringList();
Stopwatch sw = new Stopwatch().start();
int wc = 0;
for (String sentence : sentences) {
SentenceMorphParse parse = parser.parse(sentence);
wc += parse.size();
parser.disambiguate(parse);
// System.out.println(sentence);
// parse.dump();
}
System.out.println(wc);
System.out.println(sw.elapsed(TimeUnit.MILLISECONDS));
}
public Ambiguous[] getAmbiguousSequence(SentenceMorphParse sentence) {
Ambiguous[] awords = new Ambiguous[sentence.size() + 3];
awords[0] = startWord;
awords[1] = startWord;
int i = 2;
for (SentenceMorphParse.Entry entry : sentence) {
int[] roots = new int[entry.parses.size()];
int[][] igs = new int[entry.parses.size()][];
int j = 0;
for (MorphParse parse : entry.parses) {
String rootPart = parse.dictionaryItem.lemma;
roots[j] = rootLm.getVocabulary().indexOf(rootPart);
igs[j] = new int[parse.inflectionalGroups.size()];
for (int k = 0; j < parse.inflectionalGroups.size(); k++) {
igs[j][k] =
igLm.getVocabulary().indexOf(parse.inflectionalGroups.get(k).formatNoSurface());
}
j++;
}
awords[i] = new Ambiguous(roots, igs);
i++;
}
awords[i] = endWord;
return awords;
}
@Override
public void disambiguate(SentenceMorphParse sentenceParse) {
Ambiguous[] ambiguousSeq = getAmbiguousSequence(sentenceParse);
int[] bestSequence = bestSequence(ambiguousSeq);
for (int i = 0; i < bestSequence.length; i++) {
List<MorphParse> results = sentenceParse.getParses(i);
if (results.size() == 1) continue;
MorphParse tmp = results.get(0);
results.set(0, results.get(bestSequence[i]));
results.set(bestSequence[i], tmp);
}
}