public Classifier runLearner(Corpus trainCorpus, File workDir, String featSetName)
throws IOException, FileNotFoundException {
String modelName = cfg.getModelName();
String model = Utils.getWorkDirectory() + "/" + modelName;
if (modelName == null)
throw new RuntimeException("Model name needs to be specified (parameter MODEL_NAME)");
// SimpleDateFormat nameFormat = new SimpleDateFormat("yyyyMMdd");
// Date date = new Date();
// modelName = nameFormat.format(date) + "-" + modelName;
Classifier classifier = Constructor.createClassifier(model);
if (featSetName == null)
throw new RuntimeException(
"Feature set name needs to be specified (parameter FEAT_SET_NAME)");
if (modelName == null)
throw new RuntimeException("Model name needs to be specified (parameter MODEL_NAME)");
// merge feature files together
File mergedFeatureVector = File.createTempFile("mergedFeatureVector_", ".csv.gz", workDir);
OutputStream trainFeatures = new FileOutputStream(mergedFeatureVector);
FeatureMerger.combine(trainFeatures, trainCorpus);
System.out.println("start training");
classifier.train(mergedFeatureVector, new File(workDir, classifier.getName() + ".model"));
return classifier;
}
/**
* Trains a classifier using the feature files housed in the training directories in the Config.
*
* @param options - a string array of various options used in training (e.g. - where to save the
* model file, training parameters, etc.)
*/
public void train(File trainFile, File outputModelFile) {
if (mOptions == null) {
mOptions = Utils.getConfig().getStringArray("ClOptions." + getName());
}
train(trainFile, outputModelFile, mOptions);
}
