opennlp-ml/src/main/java/org/apache/opennlp/ml/maxent/ModelTrainer.java

/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 * 
 *   http://www.apache.org/licenses/LICENSE-2.0
 * 
 * Unless required by applicable law or agreed to in writing,
 * software distributed under the License is distributed on an
 * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
 * KIND, either express or implied.  See the License for the
 * specific language governing permissions and limitations
 * under the License.
 */

package org.apache.opennlp.ml.maxent;

import java.io.File;
import java.io.FileReader;

import org.apache.opennlp.ml.maxent.io.GISModelWriter;
import org.apache.opennlp.ml.maxent.io.SuffixSensitiveGISModelWriter;
import org.apache.opennlp.ml.model.AbstractModel;
import org.apache.opennlp.ml.model.AbstractModelWriter;
import org.apache.opennlp.ml.model.EventStream;
import org.apache.opennlp.ml.model.OnePassDataIndexer;
import org.apache.opennlp.ml.model.OnePassRealValueDataIndexer;
import org.apache.opennlp.ml.perceptron.PerceptronTrainer;
import org.apache.opennlp.ml.perceptron.SuffixSensitivePerceptronModelWriter;

/**
 * Main class which calls the GIS procedure after building the EventStream from
 * the data.
 */
public class ModelTrainer {

  // some parameters if you want to play around with the smoothing option
  // for model training. This can improve model accuracy, though training
  // will potentially take longer and use more memory. Model size will also
  // be larger. Initial testing indicates improvements for models built on
  // small data sets and few outcomes, but performance degradation for those
  // with large data sets and lots of outcomes.
  public static boolean USE_SMOOTHING = false;
  public static double SMOOTHING_OBSERVATION = 0.1;

  private static void usage() {
    System.err.println("java ModelTrainer [-real] dataFile modelFile");
    System.exit(1);
  }

  /**
   * Main method. Call as follows:
   * <p>
   * java ModelTrainer dataFile modelFile
   */
  public static void main(String[] args) {
    int ai = 0;
    boolean real = false;
    String type = "maxent";
    int maxit = 100;
    int cutoff = 1;
    double sigma = 1.0;

    if (args.length == 0) {
      usage();
    }
    while (args[ai].startsWith("-")) {
      if (args[ai].equals("-real")) {
        real = true;
      } else if (args[ai].equals("-perceptron")) {
        type = "perceptron";
      } else if (args[ai].equals("-maxit")) {
        maxit = Integer.parseInt(args[++ai]);
      } else if (args[ai].equals("-cutoff")) {
        cutoff = Integer.parseInt(args[++ai]);
      } else if (args[ai].equals("-sigma")) {
        sigma = Double.parseDouble(args[++ai]);
      } else {
        System.err.println("Unknown option: " + args[ai]);
        usage();
      }
      ai++;
    }
    String dataFileName = args[ai++];
    String modelFileName = args[ai];
    try {
      FileReader datafr = new FileReader(new File(dataFileName));
      EventStream es;
      if (!real) {
        es = new BasicEventStream(new PlainTextByLineDataStream(datafr), ",");
      } else {
        es = new RealBasicEventStream(new PlainTextByLineDataStream(datafr));
      }

      File outputFile = new File(modelFileName);

      AbstractModelWriter writer;

      AbstractModel model;
      if (type.equals("maxent")) {
	GIS.SMOOTHING_OBSERVATION = SMOOTHING_OBSERVATION;

        if (!real) {
          model = GIS.trainModel(es, maxit, cutoff, sigma);
        } else {
          model = GIS.trainModel(maxit, 
				 new OnePassRealValueDataIndexer(es, cutoff),              
				 USE_SMOOTHING);
        }

	writer = new SuffixSensitiveGISModelWriter(model, outputFile);

      } else if (type.equals("perceptron")) {
        //System.err.println("Perceptron training");
        model = new PerceptronTrainer().trainModel(maxit, new OnePassDataIndexer(es, cutoff), cutoff);

	writer = new SuffixSensitivePerceptronModelWriter(model, outputFile);

      } else {
        throw new RuntimeException("Unknown model type: " + type);
      }

      writer.persist();


    } catch (Exception e) {
      System.out.print("Unable to create model due to exception: ");
      System.out.println(e);
      e.printStackTrace();
    }
  }

}