edu.stanford.nlp.dcoref.sievepasses
Class DeterministicCorefSieve

java.lang.Object
  edu.stanford.nlp.dcoref.sievepasses.DeterministicCorefSieve

Direct Known Subclasses:

AliasMatch, CorefDictionaryMatch, DiscourseMatch, ExactStringMatch, LexicalChainMatch, MarkRole, PreciseConstructs, PronounMatch, RelaxedExactStringMatch, RelaxedHeadMatch, StrictHeadMatch1, StrictHeadMatch2, StrictHeadMatch3, StrictHeadMatch4

public abstract class DeterministicCorefSieve
extends java.lang.Object

Base class for a Coref Sieve. Each sieve extends this class, and set flags for its own options in the constructor.

Author:

heeyoung, mihais

Field Summary

SieveOptions

flags

Constructor Summary

DeterministicCorefSieve()
Initialize flagSet

Method Summary

boolean	checkEntityMatch(CorefCluster mentionCluster, CorefCluster potentialAntecedent, Mention mention, Mention ant, Dictionaries dict, java.util.Set<Mention> roleSet)
boolean	coreferent(Document document, CorefCluster mentionCluster, CorefCluster potentialAntecedent, Mention mention2, Mention ant, Dictionaries dict, java.util.Set<Mention> roleSet, Semantics semantics) Checks if two clusters are coreferent according to our sieve pass constraints
java.lang.String	flagsToString()
java.util.List<Mention>	getOrderedAntecedents(int antecedentSentence, int mySentence, java.util.List<Mention> orderedMentions, java.util.List<java.util.List<Mention>> orderedMentionsBySentence, Mention m1, int m1Position, java.util.Map<java.lang.Integer,CorefCluster> corefClusters, Dictionaries dict) Orders the antecedents for the given mention (m1)
void	init(java.util.Properties props)
boolean	skipThisMention(Document document, Mention m1, CorefCluster c, Dictionaries dict) Skip this mention? (search pruning)
boolean	useRoleSkip()

Methods inherited from class java.lang.Object

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

Field Detail

flags

public final SieveOptions flags

Constructor Detail

DeterministicCorefSieve

public DeterministicCorefSieve()

Initialize flagSet

Method Detail

init

public void init(java.util.Properties props)

flagsToString

public java.lang.String flagsToString()

useRoleSkip

public boolean useRoleSkip()

skipThisMention

public boolean skipThisMention(Document document,
                               Mention m1,
                               CorefCluster c,
                               Dictionaries dict)

Skip this mention? (search pruning)

checkEntityMatch

public boolean checkEntityMatch(CorefCluster mentionCluster,
                                CorefCluster potentialAntecedent,
                                Mention mention,
                                Mention ant,
                                Dictionaries dict,
                                java.util.Set<Mention> roleSet)

coreferent

public boolean coreferent(Document document,
                          CorefCluster mentionCluster,
                          CorefCluster potentialAntecedent,
                          Mention mention2,
                          Mention ant,
                          Dictionaries dict,
                          java.util.Set<Mention> roleSet,
                          Semantics semantics)
                   throws java.lang.Exception

Checks if two clusters are coreferent according to our sieve pass constraints

Parameters:

document -

Throws:

java.lang.Exception

getOrderedAntecedents

public java.util.List<Mention> getOrderedAntecedents(int antecedentSentence,
                                                     int mySentence,
                                                     java.util.List<Mention> orderedMentions,
                                                     java.util.List<java.util.List<Mention>> orderedMentionsBySentence,
                                                     Mention m1,
                                                     int m1Position,
                                                     java.util.Map<java.lang.Integer,CorefCluster> corefClusters,
                                                     Dictionaries dict)

Orders the antecedents for the given mention (m1)

Parameters:

antecedentSentence -

mySentence -

orderedMentions -

orderedMentionsBySentence -

m1 -

m1Position -

corefClusters -

dict -

Returns:

An ordering of potential antecedents depending on same/different sentence, etc.