edu.illinois.cs.cogcomp.lbjava.classify

Class DiscreteReferringFeature

java.lang.Object

edu.illinois.cs.cogcomp.lbjava.classify.Feature

edu.illinois.cs.cogcomp.lbjava.classify.DiscreteFeature

edu.illinois.cs.cogcomp.lbjava.classify.DiscreteReferrer

edu.illinois.cs.cogcomp.lbjava.classify.DiscreteReferringFeature

All Implemented Interfaces:

Serializable, Cloneable, Comparable

public class DiscreteReferringFeature
extends DiscreteReferrer

A referring discrete feature is one that has its own identifier, but whose value comes from a separate feature that it refers to.

Author:

Nick Rizzolo

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected ByteString	identifier The identifier string distinguishes this Feature from other Features.

Fields inherited from class edu.illinois.cs.cogcomp.lbjava.classify.DiscreteReferrer

referent

Fields inherited from class edu.illinois.cs.cogcomp.lbjava.classify.DiscreteFeature

BooleanValues, totalValues, valueIndex

Fields inherited from class edu.illinois.cs.cogcomp.lbjava.classify.Feature

containingPackage, generatingClassifier

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	DiscreteReferringFeature() For internal use only.
	DiscreteReferringFeature(Classifier c, ByteString i, DiscreteFeature r) Sets both the identifier and the referent.
	DiscreteReferringFeature(Classifier c, ByteString i, DiscreteFeature r, String[] av) Sets both the identifier and the referent.
	DiscreteReferringFeature(String p, String c, ByteString i, DiscreteFeature r, short vi, short t) Constructs a new referring feature.

Method Summary

Modifier and Type	Method and Description
boolean	classEquivalent(Feature f) Some features are functionally equivalent, differing only in the encoding of their values; this method will return true iff the class of this feature and f are different, but they differ only because they encode their values differently.
int	compareNameStrings(Object o) Compares only the run-time types, packages, classifier names, and identifiers of the features.
Feature	encode(String e) Returns a feature object in which any strings that are being used to represent an identifier or value have been encoded in byte strings.
boolean	equals(Object o) Two DiscreteReferringFeatures are equivalent when their containing packages, identifiers, and referent features are equivalent.
ByteString	getByteStringIdentifier() Retrieves this feature's identifier as a byte string.
Feature	getFeatureKey(Lexicon lexicon, boolean training, int label) Return the feature that should be used to index this feature into a lexicon.
double	getStrength() Returns the strength of this feature if it were to be placed in a mathematical vector space.
String	getStringIdentifier() Retrieves this feature's identifier as a string.
boolean	hasByteStringIdentifier() Determines if this feature contains a byte string identifier field.
int	hashCode() The hash code of a DiscreteReferringFeature is the sum of the hash codes of its containing package, identifier, and the referent feature.
void	lexRead(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in, Lexicon lex, String p, String g, String si, ByteString bi) Reads the representation of a feature with this object's run-time type as stored by a lexicon, overwriting the data in this object.
String	lexWrite(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out, Lexicon lex, String c, String p, String g, String si, ByteString bi) Writes a binary representation of the feature intended for use by a lexicon, omitting redundant information when possible.
RealFeature	makeReal() Returns a RealPrimitiveFeature whose value field is set to the strength of the current feature, and whose identifier field contains all the information necessary to distinguish this feature from other features.
void	read(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in) Reads the representation of a feaeture with this object's run-time type from the given stream, overwriting the data in this object.
Feature	withStrength(double s) Returns a new feature object that's identical to this feature except its strength is given by s.
void	write(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out) Writes a complete binary representation of the feature.
void	writeNameString(StringBuffer buffer) Writes a string representation of this Feature's package, generating classifier, and identifier information to the specified buffer.

Methods inherited from class edu.illinois.cs.cogcomp.lbjava.classify.DiscreteReferrer

childLexiconLookup, compareTo, depth, getByteStringValue, getReferent, getStringValue, isReferrer, removeFromChildLexicon, valueEquals, write, writeNoPackage

Methods inherited from class edu.illinois.cs.cogcomp.lbjava.classify.DiscreteFeature

conjunction, conjunctWith, getValueIndex, isDiscrete, totalValues

Methods inherited from class edu.illinois.cs.cogcomp.lbjava.classify.Feature

clone, conjunctWith, fromArray, getFeatureKey, getGeneratingClassifier, getPackage, hasStringIdentifier, isConjunctive, isPrimitive, lexReadFeature, readFeature, setArrayLength, toString, toStringNoPackage

Methods inherited from class java.lang.Object

finalize, getClass, notify, notifyAll, wait, wait, wait

Field Detail

identifier

protected ByteString identifier

The identifier string distinguishes this Feature from other Features.

Constructor Detail

DiscreteReferringFeature

protected DiscreteReferringFeature()

For internal use only.

See Also:

Feature.readFeature(ExceptionlessInputStream)

DiscreteReferringFeature

public DiscreteReferringFeature(Classifier c,
                                ByteString i,
                                DiscreteFeature r)

Sets both the identifier and the referent.

Parameters:

c - The classifier that produced this feature.

i - The new referring feature's identifier.

r - The discrete feature referred to by this new feature.

DiscreteReferringFeature

public DiscreteReferringFeature(Classifier c,
                                ByteString i,
                                DiscreteFeature r,
                                String[] av)

Sets both the identifier and the referent.

Parameters:

c - The classifier that produced this feature.

i - The new referring feature's identifier.

r - The discrete feature referred to by this new feature.

av - The allowable values of the classifier that produced r.

DiscreteReferringFeature

public DiscreteReferringFeature(String p,
                                String c,
                                ByteString i,
                                DiscreteFeature r,
                                short vi,
                                short t)

Constructs a new referring feature.

Parameters:

p - The new discrete feature's package.

c - The name of the classifier that produced this feature.

i - The new discrete feature's identifier.

r - The discrete feature referred to by this new feature.

vi - The index corresponding to the value.

t - The total allowable values for this feature.

Method Detail

hasByteStringIdentifier

public boolean hasByteStringIdentifier()

Determines if this feature contains a byte string identifier field.

Overrides:

hasByteStringIdentifier in class Feature

Returns:

true iff this feature contains a byte string identifier field.

getStringIdentifier

public String getStringIdentifier()

Retrieves this feature's identifier as a string.

Specified by:

getStringIdentifier in class Feature

Returns:

This feature's identifier as a string.

getByteStringIdentifier

public ByteString getByteStringIdentifier()

Retrieves this feature's identifier as a byte string.

Specified by:

getByteStringIdentifier in class Feature

Returns:

This feature's identifier as a byte string.

getFeatureKey

public Feature getFeatureKey(Lexicon lexicon,
                             boolean training,
                             int label)

Return the feature that should be used to index this feature into a lexicon. If it is a binary feature, we return the feature with an empty value so that the feature will be mapped to the same weight whether it is active or not. If the feature can take multiple values, then simply return the feature object as-is.

Specified by:

getFeatureKey in class Feature

Parameters:

lexicon - The lexicon into which this feature will be indexed.

training - Whether or not the learner is currently training.

label - The label of the example containing this feature, or -1 if we aren't doing per class feature counting.

Returns:

A feature object appropriate for use as the key of a map.

makeReal

public RealFeature makeReal()

Returns a RealPrimitiveFeature whose value field is set to the strength of the current feature, and whose identifier field contains all the information necessary to distinguish this feature from other features.

Specified by:

makeReal in class Feature

getStrength

public double getStrength()

Returns the strength of this feature if it were to be placed in a mathematical vector space.

Overrides:

getStrength in class DiscreteFeature

withStrength

public Feature withStrength(double s)

Returns a new feature object that's identical to this feature except its strength is given by s.

Specified by:

withStrength in class Feature

Parameters:

s - The strength of the new feature.

Returns:

A new feature object as above, or null if this feature cannot take the specified strength.

encode

public Feature encode(String e)

Returns a feature object in which any strings that are being used to represent an identifier or value have been encoded in byte strings.

Specified by:

encode in class Feature

Parameters:

e - The encoding to use.

Returns:

A feature object as above; possible this object.

hashCode

public int hashCode()

The hash code of a DiscreteReferringFeature is the sum of the hash codes of its containing package, identifier, and the referent feature.

Overrides:

hashCode in class DiscreteReferrer

Returns:

The hash code of this feature.

equals

public boolean equals(Object o)

Two DiscreteReferringFeatures are equivalent when their containing packages, identifiers, and referent features are equivalent.

Overrides:

equals in class Feature

Parameters:

o - The object with which to compare this feature.

Returns:

true iff the parameter is an equivalent feature.

classEquivalent

public boolean classEquivalent(Feature f)

Some features are functionally equivalent, differing only in the encoding of their values; this method will return true iff the class of this feature and f are different, but they differ only because they encode their values differently. This method does not compare the values themselves, however.

Overrides:

classEquivalent in class Feature

Parameters:

f - Another feature.

Returns:

See above.

compareNameStrings

public int compareNameStrings(Object o)

Compares only the run-time types, packages, classifier names, and identifiers of the features.

Overrides:

compareNameStrings in class Feature

Parameters:

o - An object to compare with.

Returns:

Integers appropriate for sorting features first by package, then by classifier name, and then by identifier.

writeNameString

public void writeNameString(StringBuffer buffer)

Writes a string representation of this Feature's package, generating classifier, and identifier information to the specified buffer.

Overrides:

writeNameString in class Feature

Parameters:

buffer - The buffer to write to.

write

public void write(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out)

Writes a complete binary representation of the feature.

Overrides:

write in class DiscreteReferrer

Parameters:

out - The output stream.

read

public void read(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in)

Reads the representation of a feaeture with this object's run-time type from the given stream, overwriting the data in this object.

Overrides:

read in class DiscreteReferrer

Parameters:

in - The input stream.

lexWrite

public String lexWrite(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out,
                       Lexicon lex,
                       String c,
                       String p,
                       String g,
                       String si,
                       ByteString bi)

Writes a binary representation of the feature intended for use by a lexicon, omitting redundant information when possible.

Overrides:

lexWrite in class DiscreteReferrer

Parameters:

out - The output stream.

lex - The lexicon out of which this feature is being written.

c - The fully qualified name of the assumed class. The runtime class of this feature won't be written if it's equivalent to c.

p - The assumed package string. This feature's package string won't be written if it's equivalent to p.

g - The assumed classifier name string. This feature's classifier name string won't be written if it's equivalent to g.

si - The assumed identifier as a string. If this feature has a string identifier, it won't be written if it's equivalent to si.

bi - The assumed identifier as a byte string. If this feature has a byte string identifier, it won't be written if it's equivalent to bi.

Returns:

The name of the runtime type of this feature.

lexRead

public void lexRead(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in,
                    Lexicon lex,
                    String p,
                    String g,
                    String si,
                    ByteString bi)

Reads the representation of a feature with this object's run-time type as stored by a lexicon, overwriting the data in this object.

This method is appropriate for reading features as written by lexWrite(ExceptionlessOutputStream,Lexicon,String,String,String,String,ByteString).

Overrides:

lexRead in class DiscreteReferrer

Parameters:

in - The input stream.

lex - The lexicon we are reading in to.

p - The assumed package string. If no package name is given in the input stream, the instantiated feature is given this package.

g - The assumed classifier name string. If no classifier name is given in the input stream, the instantiated feature is given this classifier name.

si - The assumed identifier as a string. If the feature being read has a string identifier field and no identifier is given in the input stream, the feature is given this identifier.

bi - The assumed identifier as a byte string. If the feature being read has a byte string identifier field and no identifier is given in the input stream, the feature is given this identifier.