Package opennlp.tools.ml.naivebayes

Class LogProbabilities<T>

java.lang.Object

opennlp.tools.ml.naivebayes.Probabilities<T>

opennlp.tools.ml.naivebayes.LogProbabilities<T>

Type Parameters:

T - the label (category) class

public class LogProbabilities<T>
extends Probabilities<T>

Class implementing the probability distribution over labels returned by a classifier as a log of probabilities. This is necessary because floating point precision in Java does not allow for high-accuracy representation of very low probabilities such as would occur in a text categorizer.

Constructor Summary

Constructors

Constructor	Description
LogProbabilities()

Method Summary

Modifier and Type	Method	Description
void	addIn(T t, double probability, int count)	Compounds the existing probability mass on the label with the new probability passed in to the method.
void	discardCountsBelow(double i)
Double	get(T t)	Returns the probability associated with a label
Map<T,Double>	getAll()	Returns the probabilities associated with all labels
Double	getLog(T t)	Returns the log probability associated with a label
T	getMax()	Returns the most likely label
void	set(T t, double probability)	Assigns a probability to a label, discarding any previously assigned probability.
void	set(T t, Probability<T> probability)	Assigns a probability to a label, discarding any previously assigned probability.
void	setIfLarger(T t, double probability)	Assigns a probability to a label, discarding any previously assigned probability, if the new probability is greater than the old one.
void	setLog(T t, double probability)	Assigns a log probability to a label, discarding any previously assigned probability.

Methods inherited from class opennlp.tools.ml.naivebayes.Probabilities

getConfidence, getKeys, getMaxValue, setConfidence, toString

Methods inherited from class java.lang.Object

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

Constructor Detail

LogProbabilities

public LogProbabilities()

Method Detail

set

public void set(T t,
                double probability)

Assigns a probability to a label, discarding any previously assigned probability.

Overrides:

set in class Probabilities<T>

Parameters:

t - the label to which the probability is being assigned

probability - the probability to assign

set

public void set(T t,
                Probability<T> probability)

Assigns a probability to a label, discarding any previously assigned probability.

Overrides:

set in class Probabilities<T>

Parameters:

t - the label to which the probability is being assigned

probability - the probability to assign

setIfLarger

public void setIfLarger(T t,
                        double probability)

Assigns a probability to a label, discarding any previously assigned probability, if the new probability is greater than the old one.

Overrides:

setIfLarger in class Probabilities<T>

Parameters:

t - the label to which the probability is being assigned

probability - the probability to assign

setLog

public void setLog(T t,
                   double probability)

Assigns a log probability to a label, discarding any previously assigned probability.

Overrides:

setLog in class Probabilities<T>

Parameters:

t - the label to which the log probability is being assigned

probability - the log probability to assign

addIn

public void addIn(T t,
                  double probability,
                  int count)

Compounds the existing probability mass on the label with the new probability passed in to the method.

Overrides:

addIn in class Probabilities<T>

Parameters:

t - the label whose probability mass is being updated

probability - the probability weight to add

count - the amplifying factor for the probability compounding

get

public Double get(T t)

Returns the probability associated with a label

Overrides:

get in class Probabilities<T>

Parameters:

t - the label whose probability needs to be returned

Returns:

the probability associated with the label

getLog

public Double getLog(T t)

Returns the log probability associated with a label

Overrides:

getLog in class Probabilities<T>

Parameters:

t - the label whose log probability needs to be returned

Returns:

the log probability associated with the label

discardCountsBelow

public void discardCountsBelow(double i)

Overrides:

discardCountsBelow in class Probabilities<T>

getAll

public Map<T,Double> getAll()

Returns the probabilities associated with all labels

Overrides:

getAll in class Probabilities<T>

Returns:

the HashMap of labels and their probabilities

getMax

public T getMax()

Returns the most likely label

Overrides:

getMax in class Probabilities<T>

Returns:

the label that has the highest associated probability