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;
}
public static void main(String[] args) {
try {
String corpusFile = args[0];
String goldSet = args[1];
File outputDir = new File(args[2]);
SystemConfig systemConfig = DriverUtils.configure(args);
systemConfig.setAnnotationSetName(Constants.GS_NP, goldSet);
Trainer trainer = new Trainer(systemConfig);
FeatureGenerator featureGenerator = new FeatureGenerator(systemConfig);
// get corpus
Corpus c = DriverUtils.loadFiles(corpusFile);
Preprocessor preprocessor = new Preprocessor(systemConfig);
preprocessor.preprocess(c, false);
// generate features
String featureSetName = featureGenerator.generateFeatures(c, true);
// train classifier
Classifier classifier = trainer.runLearner(c, outputDir, featureSetName);
System.out.println("classifier trained: " + classifier.getName());
} catch (IOException e) {
e.printStackTrace();
} catch (ConfigurationException e) {
e.printStackTrace();
}
}
