edu.illinois.cs.cogcomp.ner.LbjTagger

Class NEWord

java.lang.Object

edu.illinois.cs.cogcomp.lbjava.parse.LinkedChild

edu.illinois.cs.cogcomp.lbjava.nlp.Word

edu.illinois.cs.cogcomp.ner.LbjTagger.NEWord

All Implemented Interfaces:

Serializable, Cloneable

public class NEWord
extends Word

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	NEWord.DiscreteFeature
static class	NEWord.LabelToLookAt
static class	NEWord.RealFeature

Field Summary

Fields

Modifier and Type	Field and Description
String	form
ArrayList<String>	gazetteers
NamedEntity	goldEntity
boolean	isCaseNormalized
ArrayList<NEWord.RealFeature>	level1AggregationFeatures
String	neLabel This field stores the named entity type tag found in labeled data.
String	neTypeLevel1 This field is used to store a computed named entity type tag.
String	neTypeLevel2
NEWord	nextIgnoreSentenceBoundary these are referencence to previous and next words, ignoring sentence boundries.
String	normalizedForm
String	normalizedMostLinkableExpression used by wikipedia linkability.
String	originalForm
String[]	parts contains the parts of this word if DualTokenization is enabled, otherwise a waste of space.
NamedEntity	predictedEntity
CharacteristicWords	predictionConfidencesLevel1Classifier
CharacteristicWords	predictionConfidencesLevel2Classifier
NEWord	previousIgnoreSentenceBoundary
String[]	wikifierFeatures

Fields inherited from class edu.illinois.cs.cogcomp.lbjava.nlp.Word

capitalized, lemma, partOfSpeech, wordSense

Fields inherited from class edu.illinois.cs.cogcomp.lbjava.parse.LinkedChild

end, label, next, parent, previous, start

Constructor Summary

Constructors

Constructor and Description
NEWord(Word w, NEWord p, String type) An NEWord can be constructed from a Word object representing the same word, an NEWord representing the previous word in the sentence, and the named entity type label found in the data.

Method Summary

Modifier and Type	Method and Description
static void	addTokenToSentence(edu.illinois.cs.cogcomp.lbjava.parse.LinkedVector sentence, NEWord word)
static void	addTokenToSentence(edu.illinois.cs.cogcomp.lbjava.parse.LinkedVector sentence, String token, String tag) Add the provided token to the sentence, for also do any additional word spliting.
String[]	getAllNonlocalFeatures()
ArrayList<NEWord.RealFeature>	getLevel1AggregationFeatures()
HashMap<String,Integer>	getNonLocalFeatures()
int	getNonLocFeatCount(String nonLocFeat)
String	getPrediction(NEWord.LabelToLookAt labelType)
ArrayList<NEWord.RealFeature>	resetLevel1AggregationFeatures()
void	setPrediction(String label, NEWord.LabelToLookAt labelType)
String	toString() Produces a simple String representation of this word in which the neLabel field appears followed by the word's part of speech and finally the form (i.e., spelling) of the word all surrounded by parentheses.

Methods inherited from class edu.illinois.cs.cogcomp.lbjava.parse.LinkedChild

clone

Methods inherited from class java.lang.Object

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

Field Detail

neTypeLevel1

public String neTypeLevel1

This field is used to store a computed named entity type tag.

neTypeLevel2

public String neTypeLevel2

predictedEntity

public NamedEntity predictedEntity

predictionConfidencesLevel1Classifier

public CharacteristicWords predictionConfidencesLevel1Classifier

predictionConfidencesLevel2Classifier

public CharacteristicWords predictionConfidencesLevel2Classifier

goldEntity

public NamedEntity goldEntity

neLabel

public String neLabel

This field stores the named entity type tag found in labeled data.

parts

public String[] parts

contains the parts of this word if DualTokenization is enabled, otherwise a waste of space. used by the Affixes and FormParts classifiers. This appears to be an additional tokenization of what was initially split as an individual word, and results in the addition of entries in the feature vector for Affixes classifier. The dual tokenization could be performed during Affix and FormParts classification, presumable, this might result is performing this operation twice in those cases where both classifiers are utilized.

normalizedMostLinkableExpression

public String normalizedMostLinkableExpression

used by wikipedia linkability.

gazetteers

public ArrayList<String> gazetteers

wikifierFeatures

public String[] wikifierFeatures

nextIgnoreSentenceBoundary

public NEWord nextIgnoreSentenceBoundary

these are referencence to previous and next words, ignoring sentence boundries.

previousIgnoreSentenceBoundary

public NEWord previousIgnoreSentenceBoundary

level1AggregationFeatures

public ArrayList<NEWord.RealFeature> level1AggregationFeatures

form

public String form

originalForm

public String originalForm

normalizedForm

public String normalizedForm

isCaseNormalized

public boolean isCaseNormalized

Constructor Detail

NEWord

public NEWord(Word w,
              NEWord p,
              String type)

An NEWord can be constructed from a Word object representing the same word, an NEWord representing the previous word in the sentence, and the named entity type label found in the data.

Parameters:

w - Represents the same word as the NEWord being constructed.

p - The previous word in the sentence.

type - The named entity type label for this word from the data.

Method Detail

addTokenToSentence

public static void addTokenToSentence(edu.illinois.cs.cogcomp.lbjava.parse.LinkedVector sentence,
                                      String token,
                                      String tag)

Add the provided token to the sentence, for also do any additional word spliting.

Parameters:

sentence - the sentence to add the word to.

token - the individual token.

tag - the tag to annotate the word with.

addTokenToSentence

public static void addTokenToSentence(edu.illinois.cs.cogcomp.lbjava.parse.LinkedVector sentence,
                                      NEWord word)

getNonLocalFeatures

public HashMap<String,Integer> getNonLocalFeatures()

getLevel1AggregationFeatures

public ArrayList<NEWord.RealFeature> getLevel1AggregationFeatures()

resetLevel1AggregationFeatures

public ArrayList<NEWord.RealFeature> resetLevel1AggregationFeatures()

toString

public String toString()

Produces a simple String representation of this word in which the neLabel field appears followed by the word's part of speech and finally the form (i.e., spelling) of the word all surrounded by parentheses.

Overrides:

toString in class Word

Returns:

The POS bracket form of this word, or just the spelling of the word if the part of speech is not available.

getAllNonlocalFeatures

public String[] getAllNonlocalFeatures()

getNonLocFeatCount

public int getNonLocFeatCount(String nonLocFeat)

getPrediction

public String getPrediction(NEWord.LabelToLookAt labelType)

setPrediction

public void setPrediction(String label,
                          NEWord.LabelToLookAt labelType)