Examples with LabeledScoredConstituent edu.stanford.nlp.trees.LabeledScoredConstituent used on opensource projects

Example 1 with LabeledScoredConstituent

use of edu.stanford.nlp.trees.LabeledScoredConstituent in project CoreNLP by stanfordnlp.

the class NodePruner method prune.

List<Tree> prune(List<Tree> treeList, Label label, int start, int end) {
    // get reference tree
    if (treeList.size() == 1) {
        return treeList;
    }
    Tree testTree = treeList.get(0).treeFactory().newTreeNode(label, treeList);
    Tree tempTree = parser.extractBestParse(label.value(), start, end);
    // parser.restoreUnaries(tempTree);
    Tree pcfgTree = debinarizer.transformTree(tempTree);
    Set<Constituent> pcfgConstituents = pcfgTree.constituents(new LabeledScoredConstituentFactory());
    // delete child labels that are not in reference but do not cross reference
    List<Tree> prunedChildren = new ArrayList<>();
    int childStart = 0;
    for (int c = 0, numCh = testTree.numChildren(); c < numCh; c++) {
        Tree child = testTree.getChild(c);
        boolean isExtra = true;
        int childEnd = childStart + child.yield().size();
        Constituent childConstituent = new LabeledScoredConstituent(childStart, childEnd, child.label(), 0);
        if (pcfgConstituents.contains(childConstituent)) {
            isExtra = false;
        }
        if (childConstituent.crosses(pcfgConstituents)) {
            isExtra = false;
        }
        if (child.isLeaf() || child.isPreTerminal()) {
            isExtra = false;
        }
        if (pcfgTree.yield().size() != testTree.yield().size()) {
            isExtra = false;
        }
        if (!label.value().startsWith("NP^NP")) {
            isExtra = false;
        }
        if (isExtra) {
            log.info("Pruning: " + child.label() + " from " + (childStart + start) + " to " + (childEnd + start));
            log.info("Was: " + testTree + " vs " + pcfgTree);
            prunedChildren.addAll(child.getChildrenAsList());
        } else {
            prunedChildren.add(child);
        }
        childStart = childEnd;
    }
    return prunedChildren;
}