public ItemType guessLexicalItemType(String input) {
String firstWord = input.split(" ")[0];
if (firstWord.matches("[0-9]*\\-[anvr]")) return ItemType.SENSE_OFFSETS;
if (firstWord.matches("[^ ]*%[0-9]*:[^ ]*")) return ItemType.SENSE_KEYS;
if (WordNetUtils.getInstance().mapWordSenseToIWord(firstWord) != null)
return ItemType.WORD_SENSE;
if (firstWord.matches("[^ ]*#[nvra]")) return ItemType.SURFACE_TAGGED;
return ItemType.SURFACE;
}
public Pair<List<String>, List<String>> cookLexicalItem(
String text, ItemType textType, boolean discardStopwords) {
try {
List<String> cookedSentence = new ArrayList<String>();
Pair<List<String>, List<String>> out = new Pair<List<String>, List<String>>(null, null);
for (String item : Arrays.asList(text.split(" "))) {
if (item.trim().length() == 0) continue;
switch (textType) {
case SENSE_OFFSETS:
cookedSentence.add(item);
break;
case SENSE_KEYS:
IWord sense = WordNetUtils.getInstance().getSenseFromSenseKey(item);
cookedSentence.add(
GeneralUtils.fixOffset(sense.getSynset().getOffset(), sense.getPOS()));
break;
case WORD_SENSE:
IWord snse = WordNetUtils.getInstance().mapWordSenseToIWord(item);
cookedSentence.add(GeneralUtils.fixOffset(snse.getSynset().getOffset(), snse.getPOS()));
break;
case SURFACE_TAGGED:
cookedSentence.add(item);
break;
}
}
if (textType.equals(ItemType.SURFACE)) {
out = TextualSimilarity.getInstance().cookSentence(text);
cookedSentence = out.first;
}
if (cookedSentence == null) cookedSentence = new ArrayList<String>();
List<String> newCS = new ArrayList<String>();
for (String s : cookedSentence) {
// if it is a synset
if (s.matches("[0-9]*\\-[anvr]")) {
newCS.add(s);
continue;
}
String comps[] = s.split("#");
String word = comps[0];
String ps = comps[1];
// otherwise check word exists in WordNet
if (!TextualSimilarity.getInstance().isOOV(word, ps)) newCS.add(word + "#" + ps);
}
cookedSentence = newCS;
return new Pair<List<String>, List<String>>(cookedSentence, out.second);
} catch (Exception e) {
e.printStackTrace();
}
return null;
}