de.tudarmstadt.ukp.dkpro.core.stanfordnlp.util

Class StanfordAnnotator

java.lang.Object

de.tudarmstadt.ukp.dkpro.core.stanfordnlp.util.StanfordAnnotator

public class StanfordAnnotator
extends Object

A StanfordAnnotator-object creates most of the annotations for the StanfordParser component.

The code has been moved away from the Parser component because it is also used by other components (e.g. Transformations)

Constructor Summary

Constructors

Constructor and Description
StanfordAnnotator(TreeWithTokens aTokenTree)

Method Summary

Modifier and Type	Method and Description
Constituent	createConstituentAnnotation(int aBegin, int aEnd, String aConstituentType, String aSyntacticFunction) Creates a new Constituent annotation.
void	createConstituentAnnotationFromTree(edu.stanford.nlp.trees.TreebankLanguagePack aTreebankLanguagePack, boolean aCreatePos) Creates linked constituent annotations, POS annotations and lemma-annotations.
Dependency	createDependencyAnnotation(edu.stanford.nlp.trees.GrammaticalRelation aDependencyType, Token aGovernor, Token aDependent)
static Dependency	createDependencyAnnotation(org.apache.uima.jcas.JCas jCas, edu.stanford.nlp.trees.GrammaticalRelation aDependencyType, Token aGovernor, Token aDependent) Writes dependency annotations to the JCas
void	createPennTreeAnnotation(int aBegin, int aEnd) Creates annotation with Penn Treebank style representations of the syntax tree
POS	createPOSAnnotation(int aBegin, int aEnd, String aPosType) Creates a new Constituent annotation.
void	createTagAnnotation(int aBegin, int aEnd, String aTag) Creates a tag-annotation over a constituent
org.apache.uima.jcas.JCas	getJCas()
TreeWithTokens	getTokenTree()
void	recoverAnnotationsFromNodes() Recovers annotations from a Stanford Tree-Object, which have been saved within the CoreLabel of the tree.
void	setConstituentMappingProvider(MappingProvider aConstituentMappingProvider)
void	setJCas(org.apache.uima.jcas.JCas aJCas)
void	setPosMappingProvider(MappingProvider aPosMappingProvider)
void	setTokenTree(TreeWithTokens aTokenTree)

Methods inherited from class java.lang.Object

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

Constructor Detail

StanfordAnnotator

public StanfordAnnotator(TreeWithTokens aTokenTree)
                  throws org.apache.uima.cas.CASException

Throws:

org.apache.uima.cas.CASException

Method Detail

getTokenTree

public TreeWithTokens getTokenTree()

setTokenTree

public void setTokenTree(TreeWithTokens aTokenTree)

getJCas

public org.apache.uima.jcas.JCas getJCas()

setJCas

public void setJCas(org.apache.uima.jcas.JCas aJCas)

setPosMappingProvider

public void setPosMappingProvider(MappingProvider aPosMappingProvider)

setConstituentMappingProvider

public void setConstituentMappingProvider(MappingProvider aConstituentMappingProvider)

createConstituentAnnotationFromTree

public void createConstituentAnnotationFromTree(edu.stanford.nlp.trees.TreebankLanguagePack aTreebankLanguagePack,
                                                boolean aCreatePos)

Creates linked constituent annotations, POS annotations and lemma-annotations.

Note: The annotations are directly written to the indexes of the CAS.

Parameters:

aTreebankLanguagePack - the language pack.

aCreatePos - whether to create POS annotations.

createTagAnnotation

public void createTagAnnotation(int aBegin,
                                int aEnd,
                                String aTag)

Creates a tag-annotation over a constituent

Parameters:

aBegin - start-index of the constituent span

aEnd - end-index of the constituent span

aTag - the tag value

createConstituentAnnotation

public Constituent createConstituentAnnotation(int aBegin,
                                               int aEnd,
                                               String aConstituentType,
                                               String aSyntacticFunction)

Creates a new Constituent annotation. Links to parent- and child-annotations are not yet created here.

Parameters:

aBegin - start-index of the constituent span

aEnd - end-index of the constituent span

aConstituentType - the constituent type

aSyntacticFunction - the syntactic function

Returns:

the annotation

createPOSAnnotation

public POS createPOSAnnotation(int aBegin,
                               int aEnd,
                               String aPosType)

Creates a new Constituent annotation. Links to parent- and child-annotations are not yet created here.

Parameters:

aBegin - start-index of the constituent span

aEnd - end-index of the constituent span

aPosType - the constituent type

Returns:

the annotation

createDependencyAnnotation

public Dependency createDependencyAnnotation(edu.stanford.nlp.trees.GrammaticalRelation aDependencyType,
                                             Token aGovernor,
                                             Token aDependent)

createDependencyAnnotation

public static Dependency createDependencyAnnotation(org.apache.uima.jcas.JCas jCas,
                                                    edu.stanford.nlp.trees.GrammaticalRelation aDependencyType,
                                                    Token aGovernor,
                                                    Token aDependent)

Writes dependency annotations to the JCas

Parameters:

jCas - a CAS.

aDependencyType - the dependency type

aGovernor - the governing-word

aDependent - the dependent-word

Returns:

the newly created dependency annotation.

createPennTreeAnnotation

public void createPennTreeAnnotation(int aBegin,
                                     int aEnd)

Creates annotation with Penn Treebank style representations of the syntax tree

Parameters:

aBegin - start offset.

aEnd - end offset.

recoverAnnotationsFromNodes

public void recoverAnnotationsFromNodes()

Recovers annotations from a Stanford Tree-Object, which have been saved within the CoreLabel of the tree.

Note: Copying has to be done in batch, because we need to have ALL annotations that should be recovered together when copying them. The reason is that some annotations reference each other, which can cause problem if a referenced annotation has not yet been recovered.