edu.berkeley.nlp.syntax.GrammaticalRelation Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of berkeleyparser Show documentation

The Berkeley parser analyzes the grammatical structure of natural language using probabilistic context-free grammars (PCFGs).

The newest version!

package edu.berkeley.nlp.syntax;

import edu.berkeley.nlp.ling.HeadFinder;
import edu.berkeley.nlp.ling.CollinsHeadFinder;
import edu.berkeley.nlp.util.CollectionUtils;
import edu.berkeley.nlp.util.Pair;

import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;

/**
 * User: aria42
 * Date: Apr 15, 2009
 */
public class GrammaticalRelation {


  public static Pair,Tree> getSubjectObject(TreePathFinder tpf,
                                                                 Tree vpNode)
  {
    if (!vpNode.getLabel().startsWith("VP")) {
      throw new IllegalArgumentException(vpNode + " is not a VP");
    }
    List> childs = vpNode.getChildren();
    // Subject
    Tree subj = null;
    for (Tree node: tpf.getRoot().getPostOrderTraversal()) {
      if (node.getLabel().startsWith("NP")) {
        Tree lcaNode = tpf.findLowestCommonAncestor(node,vpNode);
        if (lcaNode.getLabel().startsWith("S") && tpf.findParent(node) == lcaNode) {
          subj = node;
          break;
        }
      }
    }
    // Object
    Tree obj = null;
    for (int c = 0; c < childs.size(); c++) {
      Tree child = childs.get(c);
      if (child.isPhrasal() && child.getLabel().startsWith("NP")) {
        obj = child;
        break;
      }
    }
    if (subj == null || obj == null) return null;
    return Pair.newPair(subj,obj);
  }

  private final static Set isVerbs =
    new HashSet(CollectionUtils.makeList("is","was"));
     
  public static List,Tree>> getPredicateNominativePairs(TreePathFinder tpf,
                                                                                  Tree root,
                                                                                  HeadFinder hf)
  {
    List> vpNodes = new ArrayList();
    for (Tree node : tpf.getRoot().getPostOrderTraversal()) {
      if (node.getLabel().startsWith("VP")) {
        Pair p = HeadFinder.Utils.getHeadWordAndPartOfSpeechPair(hf, node);        
        String headWord = p.getFirst();
        if (isVerbs.contains(headWord.toLowerCase())) {
          vpNodes.add(node);
        }
      }
    }
    List,Tree>> result = new ArrayList, Tree>>();
    for (Tree vpNode : vpNodes) {
      Pair, Tree> treeTreePair = getSubjectObject(tpf, vpNode);
      if (treeTreePair != null) result.add(treeTreePair);
    }
    return result;
  }

  public static void main(String[] args) {
    Tree t = Trees.PennTreeReader.parseEasy("(ROOT (S (NP (NNP John)) (VP (VBD was) (NP (DT a) (NN man)))))");
    TreePathFinder tpf = new TreePathFinder(t);
    System.out.println(getPredicateNominativePairs(tpf, t, new CollinsHeadFinder()));
  }

}