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

Class TreeWithTokens

java.lang.Object

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

public class TreeWithTokens
extends Object

A wrapper object that manages a tree object together with the respective Token annotations for the leafs of the tree. This is needed for being able to map the leaves of the tree to words in a CAS. Trees in TreeWithToken-object are always converted to trees with CoreLabel-type labels.

Constructor Summary

Constructors

Constructor and Description
TreeWithTokens(edu.stanford.nlp.trees.Tree tree, List<Token> tokens)

Method Summary

Modifier and Type	Method and Description
edu.stanford.nlp.trees.Tree	getBestFit(org.apache.uima.jcas.tcas.Annotation anno) Finds the best-fitting node in the tree for a given annotation.
edu.stanford.nlp.util.IntPair	getSpan() Returns the span of the documentText that is covered by this TreeWithTokens.
edu.stanford.nlp.util.IntPair	getSpan(edu.stanford.nlp.trees.Tree subtree) Returns the span of the documentText that is covered by a given subtree, that has to be taken directly from the original tree.
List<Token>	getTokens()
edu.stanford.nlp.trees.Tree	getTree()
void	setTokens(List<Token> tokens)
void	setTree(edu.stanford.nlp.trees.Tree tree)

Methods inherited from class java.lang.Object

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

Constructor Detail

TreeWithTokens

public TreeWithTokens(edu.stanford.nlp.trees.Tree tree,
                      List<Token> tokens)

Method Detail

setTree

public void setTree(edu.stanford.nlp.trees.Tree tree)

getTree

public edu.stanford.nlp.trees.Tree getTree()

setTokens

public void setTokens(List<Token> tokens)

getTokens

public List<Token> getTokens()

getSpan

public edu.stanford.nlp.util.IntPair getSpan()

Returns the span of the documentText that is covered by this TreeWithTokens.

Returns:

an IntPair describing the span of the documentText that is covered by this tree

getSpan

public edu.stanford.nlp.util.IntPair getSpan(edu.stanford.nlp.trees.Tree subtree)

Returns the span of the documentText that is covered by a given subtree, that has to be taken directly from the original tree.

NOTE: Possibly we could make this more general to also support general trees that are contained in the original tree, but are not directly taken from it (i.e. with different leaf-numbering). In order to do so, we would have to make a Tregex-Matching of the given subtree in the original tree to identify the positition of the given subtree.

This could be achieved by translating the subtree into a Tregex pattern and then matching this pattern against the original tree.

Parameters:

subtree - a subtree of this TreeWithTokens (it has to be a real subtree(!), because index numbering of subtree has to fit to the numbering of the original tree)

Returns:

an IntPair describing the span of the documentText that is covered by this tree

getBestFit

public edu.stanford.nlp.trees.Tree getBestFit(org.apache.uima.jcas.tcas.Annotation anno)

Finds the best-fitting node in the tree for a given annotation. The best-fitting node for an annotation is the deepest node in the tree that still completely contains the span of the given annotation. TODO Could be done more efficiently, I think. In a recursive method, for example, recursion could be stopped as soon as overlap becomes -1

Parameters:

anno - the annotation to find a best fit for

Returns:

the node of the tree that is the best fit for anno