net.yacy.cora.bayes

Class Classifier<T,K>

java.lang.Object

net.yacy.cora.bayes.Classifier<T,K>

Type Parameters:

T - A feature class

K - A category class

Direct Known Subclasses:

BayesClassifier

public abstract class Classifier<T,K>
extends java.lang.Object

Abstract base extended by any concrete classifier. It implements the basic functionality for storing categories or features and can be used to calculate basic probabilities – both category and feature probabilities. The classify function has to be implemented by the concrete classifier class.

Author:

Philipp Nolte

Field Summary

Fields

Modifier and Type	Field and Description
private java.util.Map<K,java.util.Map<T,java.lang.Integer>>	featureCountPerCategory A dictionary mapping features to their number of occurrences in each known category.
private static int	INITIAL_CATEGORY_DICTIONARY_CAPACITY Initial capacity of category dictionaries.
private static int	INITIAL_FEATURE_DICTIONARY_CAPACITY Initial capacity of feature dictionaries.
private int	memoryCapacity The initial memory capacity or how many classifications are memorized.
private java.util.Queue<Classification<T,K>>	memoryQueue The classifier's memory.
private java.util.Map<K,java.lang.Integer>	totalCategoryCount A dictionary mapping categories to their number of occurrences.
private java.util.Map<T,java.lang.Integer>	totalFeatureCount A dictionary mapping features to their number of occurrences.

Constructor Summary

Constructors

Constructor and Description
Classifier() Constructs a new classifier without any trained knowledge.

Method Summary

Methods

Modifier and Type	Method and Description
int	categoryCount(K category) Retrieves the number of occurrences of the given category.
abstract Classification<T,K>	classify(java.util.Collection<T> features) The classify method.
void	decrementCategory(K category) Decrements the count of a given category.
void	decrementFeature(T feature, K category) Decrements the count of a given feature in the given category.
int	featureCount(T feature, K category) Retrieves the number of occurrences of the given feature in the given category.
float	featureProbability(T feature, K category)
float	featureWeighedAverage(T feature, K category) Retrieves the weighed average P(feature|category) with overall weight of 1.0 and an assumed probability of 0.5.
float	featureWeighedAverage(T feature, K category, Classifier<T,K> calculator) Retrieves the weighed average P(feature|category) with overall weight of 1.0, an assumed probability of 0.5 and the given object to use for probability calculation.
float	featureWeighedAverage(T feature, K category, Classifier<T,K> calculator, float weight) Retrieves the weighed average P(feature|category) with the given weight and an assumed probability of 0.5 and the given object to use for probability calculation.
float	featureWeighedAverage(T feature, K category, Classifier<T,K> calculator, float weight, float assumedProbability) Retrieves the weighed average P(feature|category) with the given weight, the given assumed probability and the given object to use for probability calculation.
java.util.Set<K>	getCategories() Returns a Set of categories the classifier knows about.
int	getCategoriesTotal() Retrieves the total number of categories the classifier knows about.
java.util.Set<T>	getFeatures() Returns a Set of features the classifier knows about.
int	getMemoryCapacity() Retrieves the memory's capacity.
void	incrementCategory(K category) Increments the count of a given category.
void	incrementFeature(T feature, K category) Increments the count of a given feature in the given category.
void	learn(Classification<T,K> classification) Train the classifier by telling it that the given features resulted in the given category.
void	learn(K category, java.util.Collection<T> features) Train the classifier by telling it that the given features resulted in the given category.
void	reset() Resets the learned feature and category counts.
void	setMemoryCapacity(int memoryCapacity) Sets the memory's capacity.

Methods inherited from class java.lang.Object

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

Field Detail

INITIAL_CATEGORY_DICTIONARY_CAPACITY

private static final int INITIAL_CATEGORY_DICTIONARY_CAPACITY

Initial capacity of category dictionaries.

See Also:

Constant Field Values

INITIAL_FEATURE_DICTIONARY_CAPACITY

private static final int INITIAL_FEATURE_DICTIONARY_CAPACITY

Initial capacity of feature dictionaries. It should be quite big, because the features will quickly outnumber the categories.

See Also:

Constant Field Values

memoryCapacity

private int memoryCapacity

The initial memory capacity or how many classifications are memorized.

featureCountPerCategory

private java.util.Map<K,java.util.Map<T,java.lang.Integer>> featureCountPerCategory

A dictionary mapping features to their number of occurrences in each known category.

totalFeatureCount

private java.util.Map<T,java.lang.Integer> totalFeatureCount

A dictionary mapping features to their number of occurrences.

totalCategoryCount

private java.util.Map<K,java.lang.Integer> totalCategoryCount

A dictionary mapping categories to their number of occurrences.

memoryQueue

private java.util.Queue<Classification<T,K>> memoryQueue

The classifier's memory. It will forget old classifications as soon as they become too old.

Constructor Detail

Classifier

public Classifier()

Constructs a new classifier without any trained knowledge.

Method Detail

reset

public void reset()

Resets the learned feature and category counts.

getFeatures

public java.util.Set<T> getFeatures()

Returns a Set of features the classifier knows about.

Returns:

The Set of features the classifier knows about.

getCategories

public java.util.Set<K> getCategories()

Returns a Set of categories the classifier knows about.

Returns:

The Set of categories the classifier knows about.

getCategoriesTotal

public int getCategoriesTotal()

Retrieves the total number of categories the classifier knows about.

Returns:

The total category count.

getMemoryCapacity

public int getMemoryCapacity()

Retrieves the memory's capacity.

Returns:

The memory's capacity.

setMemoryCapacity

public void setMemoryCapacity(int memoryCapacity)

Sets the memory's capacity. If the new value is less than the old value, the memory will be truncated accordingly.

Parameters:

memoryCapacity - The new memory capacity.

incrementFeature

public void incrementFeature(T feature,
                    K category)

Increments the count of a given feature in the given category. This is equal to telling the classifier, that this feature has occurred in this category.

Parameters:

feature - The feature, which count to increase.

category - The category the feature occurred in.

incrementCategory

public void incrementCategory(K category)

Increments the count of a given category. This is equal to telling the classifier, that this category has occurred once more.

Parameters:

category - The category, which count to increase.

decrementFeature

public void decrementFeature(T feature,
                    K category)

Decrements the count of a given feature in the given category. This is equal to telling the classifier that this feature was classified once in the category.

Parameters:

feature - The feature to decrement the count for.

category - The category.

decrementCategory

public void decrementCategory(K category)

Decrements the count of a given category. This is equal to telling the classifier, that this category has occurred once less.

Parameters:

category - The category, which count to increase.

featureCount

public int featureCount(T feature,
               K category)

Retrieves the number of occurrences of the given feature in the given category.

Parameters:

feature - The feature, which count to retrieve.

category - The category, which the feature occurred in.

Returns:

The number of occurrences of the feature in the category.

categoryCount

public int categoryCount(K category)

Retrieves the number of occurrences of the given category.

Parameters:

category - The category, which count should be retrieved.

Returns:

The number of occurrences.

featureProbability

public float featureProbability(T feature,
                       K category)

featureWeighedAverage

public float featureWeighedAverage(T feature,
                          K category)

Retrieves the weighed average P(feature|category) with overall weight of 1.0 and an assumed probability of 0.5. The probability defaults to the overall feature probability.

Parameters:

feature - The feature, which probability to calculate.

category - The category.

Returns:

The weighed average probability.

See Also:

de.daslaboratorium.machinelearning.classifier.Classifier#featureProbability(Object, Object), de.daslaboratorium.machinelearning.classifier.Classifier#featureWeighedAverage(Object, Object, IFeatureProbability, float, float)

featureWeighedAverage

public float featureWeighedAverage(T feature,
                          K category,
                          Classifier<T,K> calculator)

Retrieves the weighed average P(feature|category) with overall weight of 1.0, an assumed probability of 0.5 and the given object to use for probability calculation.

Parameters:

feature - The feature, which probability to calculate.

category - The category.

calculator - The calculating object.

Returns:

The weighed average probability.

See Also:

de.daslaboratorium.machinelearning.classifier.Classifier#featureWeighedAverage(Object, Object, IFeatureProbability, float, float)

featureWeighedAverage

public float featureWeighedAverage(T feature,
                          K category,
                          Classifier<T,K> calculator,
                          float weight)

Retrieves the weighed average P(feature|category) with the given weight and an assumed probability of 0.5 and the given object to use for probability calculation.

Parameters:

feature - The feature, which probability to calculate.

category - The category.

calculator - The calculating object.

weight - The feature weight.

Returns:

The weighed average probability.

See Also:

de.daslaboratorium.machinelearning.classifier.Classifier#featureWeighedAverage(Object, Object, IFeatureProbability, float, float)

featureWeighedAverage

public float featureWeighedAverage(T feature,
                          K category,
                          Classifier<T,K> calculator,
                          float weight,
                          float assumedProbability)

Retrieves the weighed average P(feature|category) with the given weight, the given assumed probability and the given object to use for probability calculation.

Parameters:

feature - The feature, which probability to calculate.

category - The category.

calculator - The calculating object.

weight - The feature weight.

assumedProbability - The assumed probability.

Returns:

The weighed average probability.

learn

public void learn(K category,
         java.util.Collection<T> features)

Train the classifier by telling it that the given features resulted in the given category.

Parameters:

category - The category the features belong to.

features - The features that resulted in the given category.

learn

public void learn(Classification<T,K> classification)

Train the classifier by telling it that the given features resulted in the given category.

Parameters:

classification - The classification to learn.

classify

public abstract Classification<T,K> classify(java.util.Collection<T> features)

The classify method. It will retrieve the most likely category for the features given and depends on the concrete classifier implementation.

Parameters:

features - The features to classify.

Returns:

The category most likely.