edu.illinois.cs.cogcomp.core.datastructures.textannotation

Class TextAnnotationUtilities

java.lang.Object

edu.illinois.cs.cogcomp.core.datastructures.textannotation.TextAnnotationUtilities

public class TextAnnotationUtilities
extends Object

Author:

Vivek Srikumar

Field Summary

Fields

Modifier and Type	Field and Description
static Comparator<Constituent>	constituentEndComparator
static Comparator<Constituent>	constituentLengthComparator
static Comparator<Constituent>	constituentStartComparator
static Comparator<Constituent>	constituentStartEndComparator This comparator will sort entities on start location, but where start is equal on end as well so the shorter entities come first.
static Comparator<IntPair>	IntPairComparator
static Comparator<Sentence>	sentenceStartComparator

Constructor Summary

Constructors

Constructor and Description
TextAnnotationUtilities()

Method Summary

Modifier and Type	Method and Description
static void	copyAttributesFromTo(HasAttributes origObj, HasAttributes newObj)
static Constituent	copyConstituentWithNewTokenOffsets(TextAnnotation newTA, Constituent c, int offset) create a new constituent with token offsets shifted by the specified amount
static Relation	copyRelation(Relation r, Map<Constituent,Constituent> consMap) required: consMap *must* contain the source and target constituents for r as keys, and their values must be non-null
static void	copyViewFromTo(String vuName, TextAnnotation ta, TextAnnotation newTA, int sourceStartTokenIndex, int sourceEndTokenIndex, int offset)
static void	copyViewsFromTo(TextAnnotation ta, TextAnnotation newTA, int sourceStartTokenIndex, int sourceEndTokenIndex, int offset) copy views from the relevant span from ta to newTA.
static TextAnnotation	createFromTokenizedString(String text)
static List<String>	getSentenceList(TextAnnotation ta)
static TextAnnotation	getSubTextAnnotation(TextAnnotation ta, int sentenceId) Given a TextAnnotation object, and a sentence id, it gives a smaller TextAnnotation which contains the annotations specific to the given sentence id.
static String	getTokenSequence(TextAnnotation ta, int start, int end)
static void	mapSentenceAnnotationsToText(TextAnnotation sentenceTa, TextAnnotation textTa, int sentenceId) given a TextAnnotation for a sentence with annotations, map its annotations into a TextAnnotation object for a longer text containing that sentence.
static TextAnnotation	mapTransformedTextAnnotationToSource(TextAnnotation ta, StringTransformation st) given a TextAnnotation generated from the transformed text of a StringTransformation object, and the corresponding StringTransformation object, generate a new TextAnnotation whose annotations correspond to those of the transformed text TextAnnotation, but whose offsets correspond to the original text Example: you parse an xml-formatted news document, and use a StringTransformation to record all the places you removed xml markup or made other changes.
static void	printTextAnnotation(PrintStream out, TextAnnotation ta)

Methods inherited from class java.lang.Object

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

Field Detail

constituentStartEndComparator

public static final Comparator<Constituent> constituentStartEndComparator

This comparator will sort entities on start location, but where start is equal on end as well so the shorter entities come first.

constituentStartComparator

public static final Comparator<Constituent> constituentStartComparator

sentenceStartComparator

public static final Comparator<Sentence> sentenceStartComparator

constituentEndComparator

public static final Comparator<Constituent> constituentEndComparator

constituentLengthComparator

public static final Comparator<Constituent> constituentLengthComparator

IntPairComparator

public static final Comparator<IntPair> IntPairComparator

Constructor Detail

TextAnnotationUtilities

public TextAnnotationUtilities()

Method Detail

createFromTokenizedString

public static TextAnnotation createFromTokenizedString(String text)

getTokenSequence

public static String getTokenSequence(TextAnnotation ta,
                                      int start,
                                      int end)

getSentenceList

public static List<String> getSentenceList(TextAnnotation ta)

printTextAnnotation

public static void printTextAnnotation(PrintStream out,
                                       TextAnnotation ta)

mapSentenceAnnotationsToText

public static void mapSentenceAnnotationsToText(TextAnnotation sentenceTa,
                                                TextAnnotation textTa,
                                                int sentenceId)

given a TextAnnotation for a sentence with annotations, map its annotations into a TextAnnotation object for a longer text containing that sentence.

Parameters:

sentenceTa - annotated TextAnnotation for sentence

textTa - TextAnnotation for longer text containing sentence, without annotations for that sentence

sentenceId - index of the sentence in the longer text

getSubTextAnnotation

public static TextAnnotation getSubTextAnnotation(TextAnnotation ta,
                                                  int sentenceId)

Given a TextAnnotation object, and a sentence id, it gives a smaller TextAnnotation which contains the annotations specific to the given sentence id. The underlying text is just the sentence text, and character offsets are modified to correspond to this new shorter text.

copyViewsFromTo

public static void copyViewsFromTo(TextAnnotation ta,
                                   TextAnnotation newTA,
                                   int sourceStartTokenIndex,
                                   int sourceEndTokenIndex,
                                   int offset)

copy views from the relevant span from ta to newTA. If ta is smaller than newTA, map all constituents, changing offsets according to the value 'offset'. Otherwise, only map those constituents within the span sourceStartTokenIndex, sourceEndTokenIndex to newTA.

Parameters:

ta -

newTA -

sourceStartTokenIndex -

sourceEndTokenIndex -

offset -

copyViewFromTo

public static void copyViewFromTo(String vuName,
                                  TextAnnotation ta,
                                  TextAnnotation newTA,
                                  int sourceStartTokenIndex,
                                  int sourceEndTokenIndex,
                                  int offset)

copyRelation

public static Relation copyRelation(Relation r,
                                    Map<Constituent,Constituent> consMap)

required: consMap *must* contain the source and target constituents for r as keys, and their values must be non-null

Parameters:

r - relation to copy

consMap - map from original constituents to new counterparts

Returns:

new relation with all info copied from original, but with new source and target constituents

copyAttributesFromTo

public static void copyAttributesFromTo(HasAttributes origObj,
                                        HasAttributes newObj)

copyConstituentWithNewTokenOffsets

public static Constituent copyConstituentWithNewTokenOffsets(TextAnnotation newTA,
                                                             Constituent c,
                                                             int offset)

create a new constituent with token offsets shifted by the specified amount

Parameters:

newTA - TextAnnotation which will contain the new Constituent

c - original Constituent to copy

offset - the offset to shift token indexes of new Constituent. Can be negative.

Returns:

the new Constituent

mapTransformedTextAnnotationToSource

public static TextAnnotation mapTransformedTextAnnotationToSource(TextAnnotation ta,
                                                                  StringTransformation st)

given a TextAnnotation generated from the transformed text of a StringTransformation object, and the corresponding StringTransformation object, generate a new TextAnnotation whose annotations correspond to those of the transformed text TextAnnotation, but whose offsets correspond to the original text Example: you parse an xml-formatted news document, and use a StringTransformation to record all the places you removed xml markup or made other changes. You process the cleaned text with a set of NLP tools. This method takes the output and maps the offsets back to the xml-formatted source document. This is useful for e.g. TAC evaluations, where provenance offsets are important.

Parameters:

ta -

st -

Returns: