Examples with TregexMatcher edu.stanford.nlp.trees.tregex.TregexMatcher used on opensource projects

Example 6 with TregexMatcher

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

the class GrammaticalRelation method getRelatedNodes.

/** Given a {@code Tree} node {@code t}, attempts to
   *  return a list of nodes to which node {@code t} has this
   *  grammatical relation, with {@code t} as the governor.
   *
   *  @param t Target for finding dependents of t related by this GR
   *  @param root The root of the Tree
   *  @return A Collection of dependent nodes to which t bears this GR
   */
public Collection<TreeGraphNode> getRelatedNodes(TreeGraphNode t, TreeGraphNode root, HeadFinder headFinder) {
    Set<TreeGraphNode> nodeList = new ArraySet<>();
    for (TregexPattern p : targetPatterns) {
        // cdm: I deleted: && nodeList.isEmpty()
        // Initialize the TregexMatcher with the HeadFinder so that we
        // can use the same HeadFinder through the entire process of
        // building the dependencies
        TregexMatcher m = p.matcher(root, headFinder);
        while (m.findAt(t)) {
            TreeGraphNode target = (TreeGraphNode) m.getNode("target");
            if (target == null) {
                throw new AssertionError("Expression has no target: " + p);
            }
            nodeList.add(target);
            if (DEBUG) {
                log.info("found " + this + "(" + t + "-" + t.headWordNode() + ", " + m.getNode("target") + "-" + ((TreeGraphNode) m.getNode("target")).headWordNode() + ") using pattern " + p);
                for (String nodeName : m.getNodeNames()) {
                    if (nodeName.equals("target"))
                        continue;
                    log.info("  node " + nodeName + ": " + m.getNode(nodeName));
                }
            }
        }
    }
    return nodeList;
}

Example 7 with TregexMatcher

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

the class CorefMentionFinder method extractEnumerations.

protected static void extractEnumerations(CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    Tree tree = s.get(TreeCoreAnnotations.TreeAnnotation.class);
    SemanticGraph basicDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhancedDependency = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhancedDependency == null) {
        enhancedDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    TregexPattern tgrepPattern = enumerationsMentionPattern;
    TregexMatcher matcher = tgrepPattern.matcher(tree);
    Map<IntPair, Tree> spanToMentionSubTree = Generics.newHashMap();
    while (matcher.find()) {
        matcher.getMatch();
        Tree m1 = matcher.getNode("m1");
        Tree m2 = matcher.getNode("m2");
        List<Tree> mLeaves = m1.getLeaves();
        int beginIdx = ((CoreLabel) mLeaves.get(0).label()).get(CoreAnnotations.IndexAnnotation.class) - 1;
        int endIdx = ((CoreLabel) mLeaves.get(mLeaves.size() - 1).label()).get(CoreAnnotations.IndexAnnotation.class);
        spanToMentionSubTree.put(new IntPair(beginIdx, endIdx), m1);
        mLeaves = m2.getLeaves();
        beginIdx = ((CoreLabel) mLeaves.get(0).label()).get(CoreAnnotations.IndexAnnotation.class) - 1;
        endIdx = ((CoreLabel) mLeaves.get(mLeaves.size() - 1).label()).get(CoreAnnotations.IndexAnnotation.class);
        spanToMentionSubTree.put(new IntPair(beginIdx, endIdx), m2);
    }
    for (Map.Entry<IntPair, Tree> spanMention : spanToMentionSubTree.entrySet()) {
        IntPair span = spanMention.getKey();
        if (!mentionSpanSet.contains(span) && !insideNE(span, namedEntitySpanSet)) {
            int dummyMentionId = -1;
            Mention m = new Mention(dummyMentionId, span.get(0), span.get(1), sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(span.get(0), span.get(1))), spanMention.getValue());
            mentions.add(m);
            mentionSpanSet.add(span);
        }
    }
}

Example 8 with TregexMatcher

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

the class HybridCorefMentionFinder method extractNPorPRP.

private static void extractNPorPRP(CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    Tree tree = s.get(TreeCoreAnnotations.TreeAnnotation.class);
    tree.indexLeaves();
    SemanticGraph basicDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhancedDependency = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhancedDependency == null) {
        enhancedDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    TregexPattern tgrepPattern = npOrPrpMentionPattern;
    TregexMatcher matcher = tgrepPattern.matcher(tree);
    while (matcher.find()) {
        Tree t = matcher.getMatch();
        List<Tree> mLeaves = t.getLeaves();
        int beginIdx = ((CoreLabel) mLeaves.get(0).label()).get(CoreAnnotations.IndexAnnotation.class) - 1;
        int endIdx = ((CoreLabel) mLeaves.get(mLeaves.size() - 1).label()).get(CoreAnnotations.IndexAnnotation.class);
        // try not to have span that ends with ,
        if (",".equals(sent.get(endIdx - 1).word())) {
            endIdx--;
        }
        IntPair mSpan = new IntPair(beginIdx, endIdx);
        //      if(!mentionSpanSet.contains(mSpan) && (!insideNE(mSpan, namedEntitySpanSet)) ) {
        if (!mentionSpanSet.contains(mSpan) && (!insideNE(mSpan, namedEntitySpanSet) || t.value().startsWith("PRP"))) {
            int dummyMentionId = -1;
            Mention m = new Mention(dummyMentionId, beginIdx, endIdx, sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(beginIdx, endIdx)), t);
            mentions.add(m);
            mentionSpanSet.add(mSpan);
            if (m.originalSpan.size() > 1) {
                boolean isNE = true;
                for (CoreLabel cl : m.originalSpan) {
                    if (!cl.tag().startsWith("NNP"))
                        isNE = false;
                }
                if (isNE) {
                    namedEntitySpanSet.add(mSpan);
                }
            }
        }
    }
}

Example 9 with TregexMatcher

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

the class RuleBasedCorefMentionFinder method extractNPorPRP.

public void extractNPorPRP(CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    Tree tree = s.get(TreeCoreAnnotations.TreeAnnotation.class);
    tree.indexLeaves();
    SemanticGraph basicDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhancedDependency = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhancedDependency == null) {
        enhancedDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    TregexPattern tgrepPattern = npOrPrpMentionPattern;
    TregexMatcher matcher = tgrepPattern.matcher(tree);
    while (matcher.find()) {
        Tree t = matcher.getMatch();
        List<Tree> mLeaves = t.getLeaves();
        int beginIdx = ((CoreLabel) mLeaves.get(0).label()).get(CoreAnnotations.IndexAnnotation.class) - 1;
        int endIdx = ((CoreLabel) mLeaves.get(mLeaves.size() - 1).label()).get(CoreAnnotations.IndexAnnotation.class);
        //if (",".equals(sent.get(endIdx-1).word())) { endIdx--; } // try not to have span that ends with ,
        IntPair mSpan = new IntPair(beginIdx, endIdx);
        if (!mentionSpanSet.contains(mSpan) && (lang == Locale.CHINESE || !insideNE(mSpan, namedEntitySpanSet))) {
            //      if(!mentionSpanSet.contains(mSpan) && (!insideNE(mSpan, namedEntitySpanSet) || t.value().startsWith("PRP")) ) {
            int dummyMentionId = -1;
            Mention m = new Mention(dummyMentionId, beginIdx, endIdx, sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(beginIdx, endIdx)), t);
            mentions.add(m);
            mentionSpanSet.add(mSpan);
        //        if(m.originalSpan.size() > 1) {
        //          boolean isNE = true;
        //          for(CoreLabel cl : m.originalSpan) {
        //            if(!cl.tag().startsWith("NNP")) isNE = false;
        //          }
        //          if(isNE) {
        //            namedEntitySpanSet.add(mSpan);
        //          }
        //        }
        }
    }
}

Example 10 with TregexMatcher

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

the class MentionExtractor method findTreePattern.

private void findTreePattern(Tree tree, TregexPattern tgrepPattern, Set<Pair<Integer, Integer>> foundPairs) {
    try {
        TregexMatcher m = tgrepPattern.matcher(tree);
        while (m.find()) {
            Tree t = m.getMatch();
            Tree np1 = m.getNode("m1");
            Tree np2 = m.getNode("m2");
            Tree np3 = null;
            if (tgrepPattern.pattern().contains("m3"))
                np3 = m.getNode("m3");
            addFoundPair(np1, np2, t, foundPairs);
            if (np3 != null)
                addFoundPair(np2, np3, t, foundPairs);
        }
    } catch (Exception e) {
        // shouldn't happen....
        throw new RuntimeException(e);
    }
}