/*
Copyright (C) 2010 by
*
* Cam-Tu Nguyen
* ncamtu@ecei.tohoku.ac.jp or ncamtu@gmail.com
*
* Xuan-Hieu Phan
* pxhieu@gmail.com
*
* College of Technology, Vietnamese University, Hanoi
* Graduate School of Information Sciences, Tohoku University
*
* JVnTextPro-v.2.0 is a free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published
* by the Free Software Foundation; either version 2 of the License,
* or (at your option) any later version.
*
* JVnTextPro-v.2.0 is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with JVnTextPro-v.2.0); if not, write to the Free Software Foundation,
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
*/
package jmaxent;
import java.io.*;
import java.util.*;
// TODO: Auto-generated Javadoc
/**
* The Class Option.
*/
public class Option {
// model directory
/** The model dir. */
public String modelDir = ".";
// model file
/** The model file. */
public String modelFile = "model.txt";
/** The model separator. */
public static String modelSeparator = "##########";
// option file
/** The option file. */
public String optionFile = "option.txt";
// training data, testing data file
/** The train data file. */
public String trainDataFile = "train.labeled";
/** The test data file. */
public String testDataFile = "test.labeled";
/** The label separator. */
public static String labelSeparator = "/";
// training log
/** The train log file. */
public String trainLogFile = "trainlog.txt";
/** The is logging. */
public boolean isLogging = true;
/** The num train exps. */
public int numTrainExps = 0; // number of training examples
/** The num test exps. */
public int numTestExps = 0; // number of testing examples
/** The num labels. */
public int numLabels = 0; // number of class labels
/** The num cps. */
public int numCps = 0; // number of context predicates
/** The num features. */
public int numFeatures = 0; // number of features
// thresholds for context predicate and feature cut-off
/** The cp rare threshold. */
public int cpRareThreshold = 1;
/** The rare threshold. */
public int fRareThreshold = 1;
// training options
/** The num iterations. */
public int numIterations = 100; // number of training iterations
/** The init lambda val. */
public double initLambdaVal = 0.0; // intial value for feature weights
/** The sigma square. */
public double sigmaSquare = 100; // for smoothing
/** The eps for convergence. */
public double epsForConvergence = 0.0001; // for checking training termination
/** The m for hessian. */
public int mForHessian = 7; // for L-BFGS corrections
/** The debug level. */
public int debugLevel = 1; // control output status information
// evaluation options
/** The evaluate during training. */
public boolean evaluateDuringTraining = true; // evaluate during training
/** The save best model. */
public boolean saveBestModel = true; // save the best model with testing data
/**
* Instantiates a new option.
*/
public Option() {
}
/**
* Instantiates a new option.
*
* @param modelDir the model dir
*/
public Option(String modelDir) {
if (modelDir.endsWith(File.separator)) {
this.modelDir = modelDir.substring(0, modelDir.length() - 1);
} else {
this.modelDir = modelDir;
}
}
/**
* Read options.
*
* @return true, if successful
*/
public boolean readOptions() {
String filename = modelDir + File.separator + optionFile;
BufferedReader fin = null;
String line;
try {
fin = new BufferedReader(new FileReader(filename));
System.out.println("Reading options ...");
// read option lines
while ((line = fin.readLine()) != null) {
String trimLine = line.trim();
if (trimLine.startsWith("#")) {
// comment line
continue;
}
//System.out.println(line);
StringTokenizer strTok = new StringTokenizer(line, "= \t\r\n");
int len = strTok.countTokens();
if (len != 2) {
// invalid parameter line, ignore it
continue;
}
String strOpt = strTok.nextToken();
String strVal = strTok.nextToken();
if (strOpt.compareToIgnoreCase("trainDataFile") == 0) {
trainDataFile = strVal;
} else if (strOpt.compareToIgnoreCase("testDataFile") == 0) {
testDataFile = strVal;
} else if (strOpt.compareToIgnoreCase("isLogging") == 0) {
if (!(strVal.compareToIgnoreCase("true") == 0 ||
strVal.compareToIgnoreCase("false") == 0)) {
continue;
}
isLogging = Boolean.valueOf(strVal).booleanValue();
} else if (strOpt.compareToIgnoreCase("cpRareThreshold") == 0) {
int numTemp = Integer.parseInt(strVal);
cpRareThreshold = numTemp;
} else if (strOpt.compareToIgnoreCase("fRareThreshold") == 0) {
int numTemp = Integer.parseInt(strVal);
fRareThreshold = numTemp;
} else if (strOpt.compareToIgnoreCase("numIterations") == 0) {
int numTemp = Integer.parseInt(strVal);
numIterations = numTemp;
} else if (strOpt.compareToIgnoreCase("initLambdaVal") == 0) {
double numTemp = Double.parseDouble(strVal);
initLambdaVal = numTemp;
} else if (strOpt.compareToIgnoreCase("sigmaSquare") == 0) {
double numTemp = Double.parseDouble(strVal);
sigmaSquare = numTemp;
} else if (strOpt.compareToIgnoreCase("epsForConvergence") == 0) {
double numTemp = Double.parseDouble(strVal);
epsForConvergence = numTemp;
} else if (strOpt.compareToIgnoreCase("mForHessian") == 0) {
int numTemp = Integer.parseInt(strVal);
mForHessian = numTemp;
} else if (strOpt.compareToIgnoreCase("evaluateDuringTraining") == 0) {
if (!(strVal.compareToIgnoreCase("true") == 0 ||
strVal.compareToIgnoreCase("false") == 0)) {
continue;
}
evaluateDuringTraining = Boolean.valueOf(strVal).booleanValue();
} else if (strOpt.compareToIgnoreCase("saveBestModel") == 0) {
if (!(strVal.compareToIgnoreCase("true") == 0 ||
strVal.compareToIgnoreCase("false") == 0)) {
continue;
}
saveBestModel = Boolean.valueOf(strVal).booleanValue();
} else if (strOpt.compareToIgnoreCase("trainLogFile") == 0){
trainLogFile = strVal;
// for future use
}
else if (strOpt.compareToIgnoreCase("modelFile") == 0){
modelFile = strVal;
}
else{
//for future use
}
}
System.out.println("Reading options completed!");
} catch (IOException e) {
System.out.println(e.toString());
return false;
}
return true;
}
/**
* Open train log file.
*
* @return the prints the writer
*/
public PrintWriter openTrainLogFile() {
String filename = modelDir + File.separator + trainLogFile;
PrintWriter fout = null;
try {
fout = new PrintWriter(new OutputStreamWriter( (new FileOutputStream(filename)), "UTF-8"));
} catch (IOException e) {
System.out.println(e.toString());
return null;
}
return fout;
}
/**
* Open model file.
*
* @return the buffered reader
*/
public BufferedReader openModelFile() {
String filename = modelDir + File.separator + modelFile;
BufferedReader fin = null;
try {
fin = new BufferedReader(new InputStreamReader(new FileInputStream(filename), "UTF-8"));
} catch (IOException e) {
System.out.println(e.toString());
return null;
}
return fin;
}
/**
* Creates the model file.
*
* @return the prints the writer
*/
public PrintWriter createModelFile() {
String filename = modelDir + File.separator + modelFile;
PrintWriter fout = null;
try {
fout = new PrintWriter(new OutputStreamWriter(new FileOutputStream(filename), "UTF-8"));
} catch (IOException e) {
System.out.println(e.toString());
return null;
}
return fout;
}
/**
* Write options.
*
* @param fout the fout
*/
public void writeOptions(PrintWriter fout) {
fout.println("OPTION VALUES:");
fout.println("==============");
fout.println("Model directory: " + modelDir);
fout.println("Model file: " + modelFile);
fout.println("Option file: " + optionFile);
fout.println("Training log file: " + trainLogFile + " (this one)");
fout.println("Training data file: " + trainDataFile);
fout.println("Testing data file: " + testDataFile);
fout.println("Number of training examples " + Integer.toString(numTrainExps));
fout.println("Number of testing examples " + Integer.toString(numTestExps));
fout.println("Number of class labels: " + Integer.toString(numLabels));
fout.println("Number of context predicates: " + Integer.toString(numCps));
fout.println("Number of features: " + Integer.toString(numFeatures));
fout.println("Rare threshold for context predicates: " + Integer.toString(cpRareThreshold));
fout.println("Rare threshold for features: " + Integer.toString(fRareThreshold));
fout.println("Number of training iterations: " + Integer.toString(numIterations));
fout.println("Initial value of feature weights: " + Double.toString(initLambdaVal));
fout.println("Sigma square: " + Double.toString(sigmaSquare));
fout.println("Epsilon for convergence: " + Double.toString(epsForConvergence));
fout.println("Number of corrections in L-BFGS: " + Integer.toString(mForHessian));
if (evaluateDuringTraining) {
fout.println("Evaluation during training: true");
} else {
fout.println("Evaluation during training: false");
}
if (saveBestModel) {
fout.println("Save the best model towards testing data: true");
} else {
fout.println("Save the best model towards testing data: false");
}
fout.println();
}
} // end of class Option