private Tree<String> merge(Tree<String> leftTree, Tree<String> rightTree) {
int span = leftTree.getYield().size() + rightTree.getYield().size();
String mostFrequentLabel = spanToCategories.getCounter(span).argMax();
List<Tree<String>> children = new ArrayList<Tree<String>>();
children.add(leftTree);
children.add(rightTree);
return new Tree<String>(mostFrequentLabel, children);
}
/* Builds a lexicon from the observed tags in a list of training trees. */
public Lexicon(List<Tree<String>> trainTrees) {
for (Tree<String> trainTree : trainTrees) {
List<String> words = trainTree.getYield();
List<String> tags = trainTree.getPreTerminalYield();
for (int position = 0; position < words.size(); position++) {
String word = words.get(position);
String tag = tags.get(position);
tallyTagging(word, tag);
}
}
}
private static double testParser(Parser parser, List<Tree<String>> testTrees) {
EnglishPennTreebankParseEvaluator.LabeledConstituentEval<String> eval =
new EnglishPennTreebankParseEvaluator.LabeledConstituentEval<String>(
Collections.singleton("ROOT"),
new HashSet<String>(Arrays.asList(new String[] {"''", "``", ".", ":", ","})));
for (Tree<String> testTree : testTrees) {
List<String> testSentence = testTree.getYield();
if (testSentence.size() > MAX_LENGTH) continue;
Tree<String> guessedTree = parser.getBestParse(testSentence);
System.out.println("Guess:\n" + Trees.PennTreeRenderer.render(guessedTree));
System.out.println("Gold:\n" + Trees.PennTreeRenderer.render(testTree));
eval.evaluate(guessedTree, testTree);
}
System.out.println();
return eval.display(true);
}
