gov.sandia.cognition.learning.algorithm.delta

Class AbstractDeltaCategorizer<CategoryType>

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.learning.algorithm.delta.AbstractDeltaCategorizer<CategoryType>

Type Parameters:

CategoryType -

All Implemented Interfaces:

Evaluator<Vector,CategoryType>, Categorizer<Vector,CategoryType>, DiscriminantCategorizer<Vector,CategoryType,java.lang.Double>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

Direct Known Subclasses:

BurrowsDeltaCategorizer, CosineDeltaCategorizer

@PublicationReferences(references=@PublicationReference(author="John Burrows",title="\'Delta\': a Measure of Stylistic Difference and a Guide to Likely Authorship",type=Journal,year=2002,pages={267,287}))
public abstract class AbstractDeltaCategorizer<CategoryType>
extends AbstractCloneableSerializable
implements DiscriminantCategorizer<Vector,CategoryType,java.lang.Double>

The Burrows Delta algorithm is primarily used for authorship attribution, but can be used for other applications. This abstract class can be used to implement different variants of Burrows' Delta. The input type for this algorithm is always Vector. Each element in the vectors should correspond to a feature and all vectors should be of the same size and their elements should correspond to the same features. Each element in the vectors is expected to be the number of times the corresponding feature occurs in the text that the vector was generated from divided by the total number of features in that text. This is referred to as relative feature frequency in much of the literature. You may have to read a paper on Burrows' Delta to understand how to construct the vectors correctly. If this algorithm is going to be used for other applications the most important constraint to still obey is that all vectors should be of the same size and their elements should correspond to the same thing.

Author:

alkilli

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	AbstractDeltaCategorizer.AbstractLearner<CategoryType> Abstract learner for delta algorithms.

Field Summary

Fields

Modifier and Type	Field and Description
protected java.util.ArrayList<java.lang.Double>	featureStddev The stddev of each feature.
protected AbstractDeltaCategorizer.AbstractLearner<CategoryType>	learner The learner that was used to train this categorizer.

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	AbstractDeltaCategorizer(AbstractDeltaCategorizer.AbstractLearner<CategoryType> learner, java.util.ArrayList<java.lang.Double> featureStddev) Constructor that takes a learner and featureStddev.

Method Summary

Modifier and Type	Method and Description
abstract ValueDiscriminantPair<CategoryType,java.lang.Double>	evaluateWithDiscriminant(Vector unknownInput) This abstract method should implement evaluation aspect of this general algorithm.
java.util.Set<? extends CategoryType>	getCategories() Returns a set of all the known categories.
java.util.List<java.lang.Double>	getFeatureStddev() Getter for featureStddev

Methods inherited from class gov.sandia.cognition.util.AbstractCloneableSerializable

clone

Methods inherited from class java.lang.Object

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

Methods inherited from interface gov.sandia.cognition.evaluator.Evaluator

evaluate

Methods inherited from interface gov.sandia.cognition.util.CloneableSerializable

clone

Field Detail

learner

protected final AbstractDeltaCategorizer.AbstractLearner<CategoryType> learner

The learner that was used to train this categorizer.

featureStddev

protected final java.util.ArrayList<java.lang.Double> featureStddev

The stddev of each feature.

Constructor Detail

AbstractDeltaCategorizer

protected AbstractDeltaCategorizer(AbstractDeltaCategorizer.AbstractLearner<CategoryType> learner,
                                   java.util.ArrayList<java.lang.Double> featureStddev)

Constructor that takes a learner and featureStddev.

Parameters:

learner -

featureStddev -

Method Detail

evaluateWithDiscriminant

public abstract ValueDiscriminantPair<CategoryType,java.lang.Double> evaluateWithDiscriminant(Vector unknownInput)

This abstract method should implement evaluation aspect of this general algorithm. That is, given an unknownVector, this method should return a discriminant value paired with the corresponding most likely category. The discriminant value should be the score.

Specified by:

evaluateWithDiscriminant in interface DiscriminantCategorizer<Vector,CategoryType,java.lang.Double>

Parameters:

unknownInput -

Returns:

getCategories

public java.util.Set<? extends CategoryType> getCategories()

Returns a set of all the known categories.

Specified by:

getCategories in interface Categorizer<Vector,CategoryType>

Returns:

getFeatureStddev

public java.util.List<java.lang.Double> getFeatureStddev()

Getter for featureStddev

Returns: