SPACCC_Sentence-Splitter / src /CreateModelSS.java
espejelomar's picture
add data
393badf
raw
history blame contribute delete
No virus
3.42 kB
package SentenceSplitting;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileOutputStream;
import java.io.FileReader;
import java.io.IOException;
import java.io.Reader;
import java.nio.charset.StandardCharsets;
import opennlp.tools.dictionary.Dictionary;
import opennlp.tools.ml.EventTrainer;
import opennlp.tools.sentdetect.SentenceDetectorFactory;
import opennlp.tools.sentdetect.SentenceDetectorME;
import opennlp.tools.sentdetect.SentenceModel;
import opennlp.tools.sentdetect.SentenceSampleStream;
import opennlp.tools.util.InputStreamFactory;
import opennlp.tools.util.MarkableFileInputStreamFactory;
import opennlp.tools.util.PlainTextByLineStream;
import opennlp.tools.util.TrainingParameters;
import opennlp.tools.util.model.ModelType;
/*
* This class is used to create the sentence splitter model for Spanish clinical cases using the Apache OpenNLP API.
* Input:
* - Training file, all documents' sentences splitted, one sentence per line, everything in a single file.
* - Path where the final model will be printed.
* - Name of the created model's file.
* - File with a list of abbreviations, one abbreviation per line.
* We recommend using the abbreviations' file in order to get a better model.
* Output:
* - Sentence splitting model.
*/
public class CreateModelSS {
public static void main (String args[]) throws IOException
{
String trainFile = args[0];
String outModel = args[1];
String modelName = args[2];
String abbrFile = args[3];
// directory to save the model file that is to be generated, create this directory in prior
File destDir = new File(outModel);
// Load dictionary of abbreviations.
Dictionary abbrDictionary = makeAbbrDictionary(abbrFile);
// Load train set.
InputStreamFactory in = new MarkableFileInputStreamFactory(new File(trainFile));
// parameters used by machine learning algorithm, Maxent, to train its weights
TrainingParameters mlParams = new TrainingParameters();
mlParams.put(TrainingParameters.ITERATIONS_PARAM, Integer.toString(4000));
mlParams.put(TrainingParameters.CUTOFF_PARAM, Integer.toString(3));
mlParams.put(TrainingParameters.TRAINER_TYPE_PARAM, EventTrainer.EVENT_VALUE);
mlParams.put(TrainingParameters.ALGORITHM_PARAM, ModelType.MAXENT.name());
// Train the model.
SentenceDetectorFactory sentenceDetectorFactory = SentenceDetectorFactory.create(null, "es", true, abbrDictionary, ".?!".toCharArray());
SentenceModel sentdetectModel = SentenceDetectorME.train(
"es",
new SentenceSampleStream(new PlainTextByLineStream(in, StandardCharsets.UTF_8)),
sentenceDetectorFactory,
mlParams);
// mlParams.defaultParams());
// Print out the model into a file.
File outFile = new File(destDir,modelName);
FileOutputStream outFileStream = new FileOutputStream(outFile);
sentdetectModel.serialize(outFileStream);
}
/*
* This method reads the abbreviation list file and loads the complete list into a dictionary.
*/
public static Dictionary makeAbbrDictionary(String abbrFile) throws IOException
{
Dictionary dictionary = new Dictionary();
Reader reader = new BufferedReader(new FileReader(abbrFile));
dictionary = Dictionary.parseOneEntryPerLine(reader);
return dictionary;
}
}