MaltParser 1.4.1

org.maltparser.parser.guide.instance
Class FeatureDivideModel

java.lang.Object
  extended by org.maltparser.parser.guide.instance.FeatureDivideModel
All Implemented Interfaces:
InstanceModel, Model

public class FeatureDivideModel
extends Object
implements InstanceModel

The feature divide model is used for divide the training instances into several models according to a divide feature. Usually this strategy decrease the training and classification time, but can also decrease the accuracy of the parser.

Since:
1.0
Author:
Johan Hall

Constructor Summary
FeatureDivideModel(FeatureVector features, Model parent)
          Constructs a feature divide model.
 
Method Summary
 void addInstance(SingleDecision decision)
           
 void decreaseFrequency()
           
 FeatureVector extract()
           
 void finalizeSentence(DependencyStructure dependencyGraph)
           
 FeatureVector getDivideFeatureVector()
          Returns the divide feature vector
 int getFrequency()
          Returns the frequency (number of instances)
 ClassifierGuide getGuide()
           
 FeatureVector getMasterFeatureVector()
          Returns the "sum-up" or master feature vector
 String getModelName()
           
 Model getParent()
          Returns the parent model
 void increaseFrequency()
          Increase the frequency by 1
protected  void initSplitParam(FeatureVector featureVector)
          Initialize the feature split parameters and the split feature vector and master feature vector according to the behavior strategy.
protected  void load()
          Loads the feature divide model settings .fsm file.
 void noMoreInstances()
           
 boolean predict(SingleDecision decision)
           
 FeatureVector predictExtract(SingleDecision decision)
           
protected  void save()
          Saves the feature divide model settings .fsm file.
protected  void setFrequency(int frequency)
          Sets the frequency (number of instances)
protected  void setParent(Model parent)
          Sets the parent model
 void terminate()
           
 String toString()
           
 void train()
           
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Constructor Detail

FeatureDivideModel

public FeatureDivideModel(FeatureVector features,
                          Model parent)
                   throws MaltChainedException
Constructs a feature divide model.

Parameters:
features - the feature vector used by the atomic model.
parent - the parent guide model.
Throws:
MaltChainedException
Method Detail

addInstance

public void addInstance(SingleDecision decision)
                 throws MaltChainedException
Specified by:
addInstance in interface InstanceModel
Throws:
MaltChainedException

noMoreInstances

public void noMoreInstances()
                     throws MaltChainedException
Specified by:
noMoreInstances in interface Model
Throws:
MaltChainedException

finalizeSentence

public void finalizeSentence(DependencyStructure dependencyGraph)
                      throws MaltChainedException
Specified by:
finalizeSentence in interface Model
Throws:
MaltChainedException

predict

public boolean predict(SingleDecision decision)
                throws MaltChainedException
Specified by:
predict in interface InstanceModel
Throws:
MaltChainedException

predictExtract

public FeatureVector predictExtract(SingleDecision decision)
                             throws MaltChainedException
Specified by:
predictExtract in interface InstanceModel
Throws:
MaltChainedException

extract

public FeatureVector extract()
                      throws MaltChainedException
Specified by:
extract in interface InstanceModel
Throws:
MaltChainedException

terminate

public void terminate()
               throws MaltChainedException
Specified by:
terminate in interface Model
Throws:
MaltChainedException

train

public void train()
           throws MaltChainedException
Specified by:
train in interface InstanceModel
Throws:
MaltChainedException

initSplitParam

protected void initSplitParam(FeatureVector featureVector)
                       throws MaltChainedException
Initialize the feature split parameters and the split feature vector and master feature vector according to the behavior strategy.

Parameters:
featureVector - the parent guide model's feature vector.
Throws:
MaltChainedException

save

protected void save()
             throws MaltChainedException
Saves the feature divide model settings .fsm file.

Throws:
MaltChainedException

load

protected void load()
             throws MaltChainedException
Loads the feature divide model settings .fsm file.

Throws:
MaltChainedException

getParent

public Model getParent()
Returns the parent model

Returns:
the parent model

getGuide

public ClassifierGuide getGuide()
Specified by:
getGuide in interface Model

setParent

protected void setParent(Model parent)
                  throws MaltChainedException
Sets the parent model

Parameters:
parent - the parent model
Throws:
MaltChainedException

getModelName

public String getModelName()
                    throws MaltChainedException
Specified by:
getModelName in interface Model
Throws:
MaltChainedException

getMasterFeatureVector

public FeatureVector getMasterFeatureVector()
Returns the "sum-up" or master feature vector

Returns:
a feature vector object

getDivideFeatureVector

public FeatureVector getDivideFeatureVector()
Returns the divide feature vector

Returns:
a feature vector object

getFrequency

public int getFrequency()
Returns the frequency (number of instances)

Returns:
the frequency (number of instances)

increaseFrequency

public void increaseFrequency()
Increase the frequency by 1

Specified by:
increaseFrequency in interface InstanceModel

decreaseFrequency

public void decreaseFrequency()
Specified by:
decreaseFrequency in interface InstanceModel

setFrequency

protected void setFrequency(int frequency)
Sets the frequency (number of instances)

Parameters:
frequency - (number of instances)

toString

public String toString()
Overrides:
toString in class Object

MaltParser 1.4.1

Copyright 2007-2010 Johan Hall, Jens Nilsson and Joakim Nivre.