dragon.ml.seqmodel.feature

Interface FeatureGenerator

All Known Implementing Classes:

BasicFeatureGenerator

public interface FeatureGenerator

Interface of Feature Generator

Copyright: Copyright (c) 2005

Company: IST, Drexel University

Version:

1.0

Author:

Davis Zhou

Method Summary

Modifier and Type	Method and Description
boolean	addFeatureType(FeatureType featureType) Adds wanted feature type to the feature generator
String	getFeatureName(int featureIndex)
int	getFeatureNum()
FeatureType	getFeatureTYpe(int index) Gets the index-th feature type
int	getFeatureTypeNum() Gets the number of wanted feature types.
boolean	hasNext() Tests if next feature exists
boolean	loadFeatureData() This method will be called in testing mode.
Feature	next() Gets the next feature.
boolean	readFeatures(String featureFile) Reads feature list from a text file
boolean	saveFeatures(String featureFile) Save features to a text file
void	startScanFeaturesAt(DataSequence data, int startPos, int endPos) Starts scanning features for the specified segment
boolean	supportSegment() Tests if the feature generatore supports the segment labeling problem
boolean	train(Dataset data) This method will be called in training mode.

Method Detail

addFeatureType

boolean addFeatureType(FeatureType featureType)

Adds wanted feature type to the feature generator

Parameters:

featureType - the wanted feature type

Returns:

true if added successfully

getFeatureTypeNum

int getFeatureTypeNum()

Gets the number of wanted feature types.

Returns:

the number of wanted feature types.

getFeatureTYpe

FeatureType getFeatureTYpe(int index)

Gets the index-th feature type

Parameters:

index - the index of the feature type

Returns:

the index-th feature type

train

boolean train(Dataset data)

This method will be called in training mode. The training procedure will build a index of all possible features. It will also train each feature type if necessary.

Parameters:

data - the dataset for training

Returns:

true if trained successfully

loadFeatureData

boolean loadFeatureData()

This method will be called in testing mode. The related data for wanted feature types will be loaded. Actually this method will call the readTrainingResult method of each wanted feature type.

Returns:

true if data is loaded successfully

readFeatures

boolean readFeatures(String featureFile)

Reads feature list from a text file

Parameters:

featureFile - the name of the file containing the feature information

Returns:

true if read successfully

saveFeatures

boolean saveFeatures(String featureFile)

Save features to a text file

Parameters:

featureFile - the name of the file all features will be save to

Returns:

true if saved successfully

startScanFeaturesAt

void startScanFeaturesAt(DataSequence data,
                         int startPos,
                         int endPos)

Starts scanning features for the specified segment

Parameters:

data - the sequence data

startPos - the start position of the segment

endPos - the end position of the segment

hasNext

boolean hasNext()

Tests if next feature exists

Returns:

true if next feature exist.

next

Feature next()

Gets the next feature.

Returns:

next feature

getFeatureNum

int getFeatureNum()

Returns:

the number of unique features

getFeatureName

String getFeatureName(int featureIndex)

Parameters:

featureIndex - the index of the feature

Returns:

the name of the feature

supportSegment

boolean supportSegment()

Tests if the feature generatore supports the segment labeling problem

Returns:

true if supported