org.apache.ctakes.ytex.libsvm

Class LibSVMParser

java.lang.Object

org.apache.ctakes.ytex.kernel.BaseClassifierEvaluationParser

org.apache.ctakes.ytex.libsvm.LibSVMParser

All Implemented Interfaces:

ClassifierEvaluationParser

Direct Known Subclasses:

SVMLightParser

public class LibSVMParser
extends BaseClassifierEvaluationParser

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.ctakes.ytex.kernel.BaseClassifierEvaluationParser

BaseClassifierEvaluationParser.InstanceClassInfo

Nested classes/interfaces inherited from interface org.apache.ctakes.ytex.kernel.ClassifierEvaluationParser

ClassifierEvaluationParser.ParseOption

Field Summary

Fields

Modifier and Type	Field and Description
static Pattern	labelsPattern
static Pattern	pCost
static Pattern	pDegree
static Pattern	pGamma
static Pattern	pKernel
static Pattern	pWeight
static Pattern	totalSVPattern

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

wsDotPattern, wsPattern

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

NO, YES

Constructor Summary

Constructors

Constructor and Description
LibSVMParser()

Method Summary

Modifier and Type	Method and Description
protected SVMClassifierEvaluation	initClassifierEval(String name, String experiment, String label, String options, String instanceIdFile)
protected List<Integer>	parseClassIds(BufferedReader predictionReader) parse class ids from first line in prediction file.
void	parseDirectory(File dataDir, File outputDir) parse directory.
Integer	parseModel(String modelFile) parse svm-train model file to get the number of support vectors.
protected void	parseOptions(SVMClassifierEvaluation eval, String options)
protected void	parsePredictions(String predictionFile, String instanceFile, Properties props, String instanceIdFile, SVMClassifierEvaluation eval) parse predicted class ids, probabilities; correlate to target class ids and instance ids.
protected void	parseResults(File dataDir, File outputDir, String model, String predict, SVMClassifierEvaluation eval, Properties props) parse the results in the specified output dir.
protected void	storeResults(File dataDir, Properties props, SVMClassifierEvaluation eval) store the parsed classifier evaluation

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

checkFileRead, extractFirstToken, getClassifierEvaluationDao, getFileBaseName, initClassifierEvaluation, initClassifierEvaluationFromProperties, loadClassIdMap, loadClassInfo, loadInstanceClassInfo, loadProps, parseDoubleOption, parseInstanceIds, parseIntOption, setClassifierEvaluationDao, storeSemiSupervised, updateSemiSupervisedPredictions, updateSemiSupervisedPredictions

Methods inherited from class java.lang.Object

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

Field Detail

labelsPattern

public static Pattern labelsPattern

totalSVPattern

public static Pattern totalSVPattern

pKernel

public static Pattern pKernel

pGamma

public static Pattern pGamma

pCost

public static Pattern pCost

pWeight

public static Pattern pWeight

pDegree

public static Pattern pDegree

Constructor Detail

LibSVMParser

public LibSVMParser()

Method Detail

parseModel

public Integer parseModel(String modelFile)
                   throws IOException

parse svm-train model file to get the number of support vectors. Needed for model selection

Parameters:

modelFile -

Returns:

Throws:

IOException

parseClassIds

protected List<Integer> parseClassIds(BufferedReader predictionReader)
                               throws IOException

parse class ids from first line in prediction file. this correspond to probabilities

Parameters:

predictionReader -

Returns:

Throws:

IOException

initClassifierEval

protected SVMClassifierEvaluation initClassifierEval(String name,
                                                     String experiment,
                                                     String label,
                                                     String options,
                                                     String instanceIdFile)

parsePredictions

protected void parsePredictions(String predictionFile,
                                String instanceFile,
                                Properties props,
                                String instanceIdFile,
                                SVMClassifierEvaluation eval)
                         throws IOException

parse predicted class ids, probabilities; correlate to target class ids and instance ids.

Parameters:

predictionFile - prediction (output)

instanceFile - input data file; contains target class ids

props -

instanceIdFile - instance ids corresponding to lines in input data file

eval -

Throws:

IOException

parseOptions

protected void parseOptions(SVMClassifierEvaluation eval,
                            String options)

parseDirectory

public void parseDirectory(File dataDir,
                           File outputDir)
                    throws IOException

parse directory. Expect following files:

• model.txt - libsvm model file
• options.properties - properties file with needed parameter settings (see ParseOption)
• predict.txt - predictions on test set

Throws:

IOException

storeResults

protected void storeResults(File dataDir,
                            Properties props,
                            SVMClassifierEvaluation eval)
                     throws IOException

store the parsed classifier evaluation

Parameters:

props -

eval -

Throws:

IOException

parseResults

protected void parseResults(File dataDir,
                            File outputDir,
                            String model,
                            String predict,
                            SVMClassifierEvaluation eval,
                            Properties props)
                     throws IOException

parse the results in the specified output dir. use reference data from dataDir.

Parameters:

dataDir -

outputDir -

model -

predict -

eval -

props -

Throws:

IOException