org.apache.ctakes.ytex.semil

Class SemiLFormatterFactory.SemiLDataFormatter

java.lang.Object

org.apache.ctakes.ytex.kernel.BaseSparseDataFormatter

org.apache.ctakes.ytex.libsvm.LibSVMFormatterFactory.LibSVMFormatter

org.apache.ctakes.ytex.svmlight.SVMLightFormatterFactory.SVMLightFormatter

org.apache.ctakes.ytex.semil.SemiLFormatterFactory.SemiLDataFormatter

All Implemented Interfaces:

SparseDataFormatter

Direct Known Subclasses:

SGTFormatterFactory.SGTFormatter, SVMLinFormatterFactory.SVMLinDataFormatter

Enclosing class:

SemiLFormatterFactory

public static class SemiLFormatterFactory.SemiLDataFormatter
extends SVMLightFormatterFactory.SVMLightFormatter

Field Summary

Fields

Modifier and Type	Field and Description
protected InstanceData	instanceLabel

Fields inherited from class org.apache.ctakes.ytex.svmlight.SVMLightFormatterFactory.SVMLightFormatter

foldInstanceLabelMap

Fields inherited from class org.apache.ctakes.ytex.kernel.BaseSparseDataFormatter

exportProperties, kernelUtil, labelToClassIndexMap, maxAttributeIndex, nominalAttributeMap, numericAttributeMap, outdir

Fields inherited from interface org.apache.ctakes.ytex.kernel.SparseDataFormatter

ATTR_INSTANCE_ID, SCOPE, SCOPE_FOLD, SCOPE_LABEL

Constructor Summary

Constructors

Constructor and Description
SemiLDataFormatter(KernelUtil kernelUtil)

Method Summary

Modifier and Type	Method and Description
protected void	exportData(SparseData sparseData, String label, Integer run, Integer fold)
void	exportFold(SparseData sparseData, SortedMap<Long,String> instanceClassMap, boolean train, String label, Integer run, Integer fold) export the given train/test set
protected void	exportLabel(String lblFilename, SortedMap<Long,Integer> mapInstanceIdToClass) Write the 'label' file.
protected void	fillLabelToClassIndexMap(Map<String,SortedSet<String>> labelToClassMap)
protected SortedMap<Long,Integer>	getTrainingClassMap(String idFilename, SortedMap<Long,String> trainInstanceClassMap, SortedMap<Long,String> testInstanceClassMap, Map<String,Integer> classToIndexMap, SortedSet<Long> instanceIds) pick through the training and test sets, figure out the class id for all instance ids for training.
void	initializeExport(InstanceData instanceLabel, Properties properties, SparseData sparseData) write distance up to 6 digit precision.
void	initializeFold(SparseData sparseData, String label, Integer run, Integer fold, SortedMap<Boolean,SortedMap<Long,String>> foldInstanceLabelMap) write a file with the attribute names corresponding to the indices in the libsvm data file
void	initializeLabel(String label, SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>> labelInstances, Properties properties, SparseData sparseData)

Methods inherited from class org.apache.ctakes.ytex.svmlight.SVMLightFormatterFactory.SVMLightFormatter

clearFold, exportTransductiveData

Methods inherited from class org.apache.ctakes.ytex.libsvm.LibSVMFormatterFactory.LibSVMFormatter

clearLabel, exportDataForInstances, exportDataForLabel, writeLibsvmLine

Methods inherited from class org.apache.ctakes.ytex.kernel.BaseSparseDataFormatter

addNumericAttribute, exportAttributeNames, exportAttributeNames, exportSparseMatrix, exportSparseRow, getSparseLineValues, updateLabelClassMapTransductive

Methods inherited from class java.lang.Object

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

Field Detail

instanceLabel

protected InstanceData instanceLabel

Constructor Detail

SemiLDataFormatter

public SemiLDataFormatter(KernelUtil kernelUtil)

Method Detail

exportData

protected void exportData(SparseData sparseData,
                          String label,
                          Integer run,
                          Integer fold)
                   throws IOException

Parameters:

foldInstanceLabelMap -

sparseData -

label -

run -

fold -

Throws:

IOException

exportFold

public void exportFold(SparseData sparseData,
                       SortedMap<Long,String> instanceClassMap,
                       boolean train,
                       String label,
                       Integer run,
                       Integer fold)
                throws IOException

Description copied from class: SVMLightFormatterFactory.SVMLightFormatter

export the given train/test set

Specified by:

exportFold in interface SparseDataFormatter

Overrides:

exportFold in class SVMLightFormatterFactory.SVMLightFormatter

Throws:

IOException

exportLabel

protected void exportLabel(String lblFilename,
                           SortedMap<Long,Integer> mapInstanceIdToClass)
                    throws IOException

Write the 'label' file. This contains the training labels (label != 0) and test labels (labeled as 0) and unlabeled data (labeled as 0). The order of the label file corresponds to the order in the data file.

Parameters:

lblFilename - filename to write to

mapInstanceIdToClass - sorted map of instance id to class. this must correspond to the order in the data file

Throws:

IOException

getTrainingClassMap

protected SortedMap<Long,Integer> getTrainingClassMap(String idFilename,
                                                      SortedMap<Long,String> trainInstanceClassMap,
                                                      SortedMap<Long,String> testInstanceClassMap,
                                                      Map<String,Integer> classToIndexMap,
                                                      SortedSet<Long> instanceIds)
                                               throws IOException

pick through the training and test sets, figure out the class id for all instance ids for training. If the instance is in the training set and labeled it will get the appropriate class label. If the instance is in the test set and labeled it will be unlabeled. If the instance is in the training and test sets, it will use whatever label was specified for the training set. Unlabeled instances are given the class id 0.

Write the [prefix]class.txt file

Parameters:

idFilename - filename to write instance id\ttrain/test flag\ttarget class to

trainInstanceClassMap - instances for training

testInstanceClassMap - instance for testing

classToIndexMap - map of class to class ids

instanceIds - sorted set of instance ids; the order with which class.txt will be written, and the order with which instances will appear in the training data file.

Returns:

map of instance id to class id for training

Throws:

IOException

initializeExport

public void initializeExport(InstanceData instanceLabel,
                             Properties properties,
                             SparseData sparseData)
                      throws IOException

write distance up to 6 digit precision. only write distance if < 0.999. format: row column dist 1-based indices.

Specified by:

initializeExport in interface SparseDataFormatter

Overrides:

initializeExport in class SVMLightFormatterFactory.SVMLightFormatter

Parameters:

data -

wData -

Throws:

IOException

initializeFold

public void initializeFold(SparseData sparseData,
                           String label,
                           Integer run,
                           Integer fold,
                           SortedMap<Boolean,SortedMap<Long,String>> foldInstanceLabelMap)
                    throws IOException

Description copied from class: LibSVMFormatterFactory.LibSVMFormatter

write a file with the attribute names corresponding to the indices in the libsvm data file

Specified by:

initializeFold in interface SparseDataFormatter

Overrides:

initializeFold in class SVMLightFormatterFactory.SVMLightFormatter

Throws:

IOException

initializeLabel

public void initializeLabel(String label,
                            SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>> labelInstances,
                            Properties properties,
                            SparseData sparseData)
                     throws IOException

Specified by:

initializeLabel in interface SparseDataFormatter

Overrides:

initializeLabel in class LibSVMFormatterFactory.LibSVMFormatter

Throws:

IOException

fillLabelToClassIndexMap

protected void fillLabelToClassIndexMap(Map<String,SortedSet<String>> labelToClassMap)

Overrides:

fillLabelToClassIndexMap in class SVMLightFormatterFactory.SVMLightFormatter