Пример #1
0
  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();
    }
  }