private static List<TaggedWord> cleanTags(List twList, TreebankLanguagePack tlp) {
int sz = twList.size();
List<TaggedWord> l = new ArrayList<TaggedWord>(sz);
for (int i = 0; i < sz; i++) {
TaggedWord tw = (TaggedWord) twList.get(i);
TaggedWord tw2 = new TaggedWord(tw.word(), tlp.basicCategory(tw.tag()));
l.add(tw2);
}
return l;
}
public ArrayList<String> getNounsFromSentence(String sentence) {
ArrayList<TaggedWord> tw = parseSentenceTD(sentence);
ArrayList<String> nouns = new ArrayList<String>();
for (TaggedWord t : tw) {
if (t.tag().startsWith("N")) {
nouns.add(t.value());
}
}
return nouns;
}
@Override
public void train(List<TaggedWord> sentence) {
lex.train(sentence, 1.0);
String last = null;
for (TaggedWord tagLabel : sentence) {
String tag = tagLabel.tag();
tagIndex.add(tag);
if (last == null) {
initial.incrementCount(tag);
} else {
ruleCounter.incrementCount2D(last, tag);
}
last = tag;
}
}
/**
* Turns a sentence into a flat phrasal tree. The structure is S -> tag*. And then each tag goes
* to a word. The tag is either found from the label or made "WD". The tag and phrasal node have a
* StringLabel.
*
* @param s The Sentence to make the Tree from
* @param lf The LabelFactory with which to create the new Tree labels
* @return The one phrasal level Tree
*/
public static Tree toFlatTree(Sentence<?> s, LabelFactory lf) {
List<Tree> daughters = new ArrayList<Tree>(s.length());
for (HasWord word : s) {
Tree wordNode = new LabeledScoredTreeLeaf(lf.newLabel(word.word()));
if (word instanceof TaggedWord) {
TaggedWord taggedWord = (TaggedWord) word;
wordNode =
new LabeledScoredTreeNode(
new StringLabel(taggedWord.tag()), Collections.singletonList(wordNode));
} else {
wordNode =
new LabeledScoredTreeNode(lf.newLabel("WD"), Collections.singletonList(wordNode));
}
daughters.add(wordNode);
}
return new LabeledScoredTreeNode(new StringLabel("S"), daughters);
}