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

Class TreeUtils

java.lang.Object

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

public class TreeUtils
extends Object

Utility class for the StanfordParser

Constructor Summary

Constructors

Constructor and Description
TreeUtils()

Method Summary

Modifier and Type	Method and Description
static edu.stanford.nlp.trees.Tree	createStanfordTree(org.apache.uima.jcas.tcas.Annotation root) Recursively creates an edu.stanford.nlp.trees.Tree from a ROOT annotation It also saves the whitespaces before and after a token as CoreAnnotation.BeforeAnnotation and CoreAnnotation.AfterAnnotation in the respective label of the current node.
static edu.stanford.nlp.trees.Tree	createStanfordTree(org.apache.uima.jcas.tcas.Annotation root, edu.stanford.nlp.trees.TreeFactory tFact)
static edu.stanford.nlp.trees.Tree	createStanfordTreeWithAnnotations(org.apache.uima.jcas.tcas.Annotation root) Recreates a Stanford Tree from the StanfordParser annotations and saves all non-StanfordParser-Annotations within the scope of the sentence in the label of the best fitting node.
static List<Token>	getTokenListFromTree(org.apache.uima.jcas.JCas aJCas, edu.stanford.nlp.trees.Tree t) Returns a list of Token annotations from a Tree-object
static edu.stanford.nlp.trees.Tree	pennString2Tree(String pennString) Reads in a Penn Treebank-style String and returns a tree.
static String	pennString2Words(String penn) Returns the sentence from its tree representation.
static void	reIndexLeaves(edu.stanford.nlp.trees.Tree t) Reimplementation of the indexLeaves() method of stanford tree objects.
static String	tree2Words(edu.stanford.nlp.trees.Tree t) Returns the sentence from its tree representation.

Methods inherited from class java.lang.Object

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

Constructor Detail

TreeUtils

public TreeUtils()

Method Detail

createStanfordTree

public static edu.stanford.nlp.trees.Tree createStanfordTree(org.apache.uima.jcas.tcas.Annotation root)

Recursively creates an edu.stanford.nlp.trees.Tree from a ROOT annotation It also saves the whitespaces before and after a token as CoreAnnotation.BeforeAnnotation and CoreAnnotation.AfterAnnotation in the respective label of the current node.

Parameters:

root - the ROOT annotation

Returns:

an Tree object representing the syntax structure of the sentence

createStanfordTree

public static edu.stanford.nlp.trees.Tree createStanfordTree(org.apache.uima.jcas.tcas.Annotation root,
                                                             edu.stanford.nlp.trees.TreeFactory tFact)

createStanfordTreeWithAnnotations

public static edu.stanford.nlp.trees.Tree createStanfordTreeWithAnnotations(org.apache.uima.jcas.tcas.Annotation root)
                                                                     throws org.apache.uima.cas.CASException

Recreates a Stanford Tree from the StanfordParser annotations and saves all non-StanfordParser-Annotations within the scope of the sentence in the label of the best fitting node.

CAUTION: This method is intended for the use by CAS Multipliers, which create new CASes from this tree. The annotation-spans in the source-CAS will be changed!!!!!! You do NOT want to use the source CAS after this method has been called. The createStanfordTree()-method does not change the CAS, so use this instead, if the annotations do not have to be recovered or accessed in the tree.

TODO: This behavior could be changed by making COPIES of the annotations and changing the copied instead of the originals. However, in order to being able to make copies, a dummy CAS must be introduced to which the annotations can be copied. When they are recovered, they will be copied to the new destination CAS anyway.

Parameters:

root - the ROOT annotation

Returns:

an Tree object representing the syntax structure of the sentence

Throws:

org.apache.uima.cas.CASException - if the JCas cannot be accessed.

pennString2Words

public static String pennString2Words(String penn)

Returns the sentence from its tree representation.

Parameters:

penn - the tree representation of the sentence

Returns:

the sentence

tree2Words

public static String tree2Words(edu.stanford.nlp.trees.Tree t)

Returns the sentence from its tree representation.

Parameters:

t - the tree representation of the sentence

Returns:

the sentence

getTokenListFromTree

public static List<Token> getTokenListFromTree(org.apache.uima.jcas.JCas aJCas,
                                               edu.stanford.nlp.trees.Tree t)

Returns a list of Token annotations from a Tree-object

Parameters:

aJCas - a JCas.

t - a tree.

Returns:

the tokens.

reIndexLeaves

public static void reIndexLeaves(edu.stanford.nlp.trees.Tree t)

Reimplementation of the indexLeaves() method of stanford tree objects. This method reindexes already indexed trees starting with index 1. The method expects trees with CoreMap-type labels.

Parameters:

t - a tree with CoreLabel-type labels.

See Also:

Tree.indexLeaves()

pennString2Tree

public static edu.stanford.nlp.trees.Tree pennString2Tree(String pennString)

Reads in a Penn Treebank-style String and returns a tree.

Parameters:

pennString - A Penn Treebank-style String as produced by the StandfordParser

Returns:

a tree representation of the PennString (LabeledScoredTree)