edu.stanford.nlp.ling.tokensregex.CoreMapNodePatternTrigger Maven / Gradle / Ivy

Show more of this group Show more artifacts with this name
Show all versions of stanford-corenlp Show documentation

Stanford CoreNLP provides a set of natural language analysis tools which can take raw English language text input and give the base forms of words, their parts of speech, whether they are names of companies, people, etc., normalize dates, times, and numeric quantities, mark up the structure of sentences in terms of phrases and word dependencies, and indicate which noun phrases refer to the same entities. It provides the foundational building blocks for higher level text understanding applications.

There is a newer version: 4.5.7

Show newest version

package edu.stanford.nlp.ling.tokensregex;

import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.util.*;

import java.util.*;
import java.util.function.Function;

/**
* Trigger for CoreMap Node Patterns.  Allows for fast identification of which patterns
*  may match for one node.
*
* @author Angel Chang
*/
public class CoreMapNodePatternTrigger implements MultiPatternMatcher.NodePatternTrigger {
  Collection> patterns;
  Collection> alwaysTriggered = new LinkedHashSet<>();
  TwoDimensionalCollectionValuedMap> annotationTriggers =
          new TwoDimensionalCollectionValuedMap<>();
  TwoDimensionalCollectionValuedMap> lowercaseStringTriggers =
          new TwoDimensionalCollectionValuedMap<>();

  public CoreMapNodePatternTrigger(SequencePattern... patterns) {
    this(Arrays.asList(patterns));
  }
  public CoreMapNodePatternTrigger(Collection> patterns) {
    this.patterns = patterns;

    Function, StringTriggerCandidate> stringTriggerFilter =
        in -> {
          if (in instanceof CoreMapNodePattern) {
            CoreMapNodePattern p = (CoreMapNodePattern) in;
            for (Pair v:p.getAnnotationPatterns()) {
              if (v.second instanceof CoreMapNodePattern.StringAnnotationPattern) {
                return new StringTriggerCandidate(v.first, ((CoreMapNodePattern.StringAnnotationPattern) v.second).target,
                        ((CoreMapNodePattern.StringAnnotationPattern) v.second).ignoreCase());
              }
            }
          }
          return null;
        };

    for (SequencePattern pattern:patterns) {
      // Look for first string...
      Collection triggerCandidates = pattern.findNodePatterns(stringTriggerFilter, false, true);
      // TODO: Select most unlikely to trigger trigger from the triggerCandidates
      //  (if we had some statistics on most frequent annotation values...., then pick least frequent)
      // For now, just pick the longest: going from (text or lemma) to rest
      StringTriggerCandidate trigger = triggerCandidates.stream().max(STRING_TRIGGER_CANDIDATE_COMPARATOR).orElse(null);
      if (!triggerCandidates.isEmpty()) {
        if (trigger.ignoreCase) {
          lowercaseStringTriggers.add(trigger.key, trigger.value.toLowerCase(), pattern);
        } else {
          annotationTriggers.add(trigger.key, trigger.value, pattern);
        }
      } else {
        alwaysTriggered.add(pattern);
      }
    }
  }

  private static class StringTriggerCandidate {
    Class key;
    String value;
    boolean ignoreCase;
    int keyLevel;
    int effectiveValueLength;

    public StringTriggerCandidate(Class key, String value, boolean ignoreCase) {
      this.key = key;
      this.value = value;
      this.ignoreCase = ignoreCase;
      // Favor text and lemma (more likely to be unique)
      this.keyLevel = (CoreAnnotations.TextAnnotation.class.equals(key) || CoreAnnotations.LemmaAnnotation.class.equals(key))? 1:0;
      // Special case for -LRB- ( and -RRB- )
      this.effectiveValueLength = ("-LRB-".equals(value) || "-RRB-".equals(value))? 1: value.length();
    }
  }
  private static final Comparator STRING_TRIGGER_CANDIDATE_COMPARATOR =
    new Comparator() {
      @Override
      public int compare(StringTriggerCandidate o1, StringTriggerCandidate o2) {
        if (o1.keyLevel != o2.keyLevel) {
          return (o1.keyLevel < o2.keyLevel)? -1:1;
        } else {
          int v1 = o1.effectiveValueLength;
          int v2 = o2.effectiveValueLength;
          if (v1 != v2) return (v1 < v2)? -1:1;
          else return 0;
        }
      }
    };

  @Override
  public Collection> apply(CoreMap in) {
    Set> triggeredPatterns = new LinkedHashSet<>();
    triggeredPatterns.addAll(alwaysTriggered);
    for (Class key:annotationTriggers.firstKeySet()) {
      Object value = in.get(key);
      if (value != null) {
        Collection> triggered = annotationTriggers.get(key, value);
        if (triggered != null) {
          triggeredPatterns.addAll(triggered);
        }
      }
    }
    for (Class key:lowercaseStringTriggers.firstKeySet()) {
      Object value = in.get(key);
      if (value != null && value instanceof String) {
        Collection> triggered = lowercaseStringTriggers.get(key, ((String) value).toLowerCase());
        if (triggered != null) {
          triggeredPatterns.addAll(triggered);
        }
      }
    }
    // TODO: triggers for normalized patterns...
    return triggeredPatterns;
  }
}