edu.stanford.nlp.dcoref
Class RuleBasedCorefMentionFinder

java.lang.Object
  edu.stanford.nlp.dcoref.RuleBasedCorefMentionFinder

All Implemented Interfaces:

CorefMentionFinder

public class RuleBasedCorefMentionFinder
extends java.lang.Object
implements CorefMentionFinder

Field Summary

protected boolean	assignIds
protected int	maxID
protected Annotator	parserProcessor

Constructor Summary

RuleBasedCorefMentionFinder()

Method Summary

protected void	extractEnumerations(CoreMap s, java.util.List<Mention> mentions, java.util.Set<IntPair> mentionSpanSet, java.util.Set<IntPair> namedEntitySpanSet) Extract enumerations (A, B, and C)
protected void	extractNamedEntityMentions(CoreMap s, java.util.List<Mention> mentions, java.util.Set<IntPair> mentionSpanSet, java.util.Set<IntPair> namedEntitySpanSet)
protected void	extractNPorPRP(CoreMap s, java.util.List<Mention> mentions, java.util.Set<IntPair> mentionSpanSet, java.util.Set<IntPair> namedEntitySpanSet)
java.util.List<java.util.List<Mention>>	extractPredictedMentions(Annotation doc, int _maxID, Dictionaries dict) Main method of mention detection.
java.util.List<java.util.List<Mention>>	filterPredictedMentions(java.util.List<java.util.List<Mention>> allGoldMentions, Annotation doc, Dictionaries dict) When mention boundaries are given
protected void	findHead(CoreMap s, java.util.List<Mention> mentions)
protected Tree	findSyntacticHead(Mention m, Tree root, java.util.List<CoreLabel> tokens)
protected static void	removeSpuriousMentions(CoreMap s, java.util.List<Mention> mentions, Dictionaries dict) Filter out all spurious mentions
protected static void	setBarePlural(java.util.List<Mention> mentions)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

assignIds

protected boolean assignIds

maxID

protected int maxID

parserProcessor

protected Annotator parserProcessor

Constructor Detail

RuleBasedCorefMentionFinder

public RuleBasedCorefMentionFinder()

Method Detail

filterPredictedMentions

public java.util.List<java.util.List<Mention>> filterPredictedMentions(java.util.List<java.util.List<Mention>> allGoldMentions,
                                                                       Annotation doc,
                                                                       Dictionaries dict)

When mention boundaries are given

extractPredictedMentions

public java.util.List<java.util.List<Mention>> extractPredictedMentions(Annotation doc,
                                                                        int _maxID,
                                                                        Dictionaries dict)

Main method of mention detection. Extract all NP, PRP or NE, and filter out by manually written patterns.

Specified by:

extractPredictedMentions in interface CorefMentionFinder

Parameters:

doc - The syntactically annotated document

_maxID - The last mention ID assigned. New ones are assigned starting one above this number.

dict - Dictionaries for coref.

Returns:

For each of the List of sentences in the document, a List of Mention objects

setBarePlural

protected static void setBarePlural(java.util.List<Mention> mentions)

extractNamedEntityMentions

protected void extractNamedEntityMentions(CoreMap s,
                                          java.util.List<Mention> mentions,
                                          java.util.Set<IntPair> mentionSpanSet,
                                          java.util.Set<IntPair> namedEntitySpanSet)

extractNPorPRP

protected void extractNPorPRP(CoreMap s,
                              java.util.List<Mention> mentions,
                              java.util.Set<IntPair> mentionSpanSet,
                              java.util.Set<IntPair> namedEntitySpanSet)

extractEnumerations

protected void extractEnumerations(CoreMap s,
                                   java.util.List<Mention> mentions,
                                   java.util.Set<IntPair> mentionSpanSet,
                                   java.util.Set<IntPair> namedEntitySpanSet)

Extract enumerations (A, B, and C)

findHead

protected void findHead(CoreMap s,
                        java.util.List<Mention> mentions)

findSyntacticHead

protected Tree findSyntacticHead(Mention m,
                                 Tree root,
                                 java.util.List<CoreLabel> tokens)

removeSpuriousMentions

protected static void removeSpuriousMentions(CoreMap s,
                                             java.util.List<Mention> mentions,
                                             Dictionaries dict)

Filter out all spurious mentions