Package opennlp.tools.ml.maxent

Klasse GISTrainer

java.lang.Object
opennlp.tools.ml.AbstractTrainer
opennlp.tools.ml.AbstractEventTrainer
opennlp.tools.ml.maxent.GISTrainer
Alle implementierten Schnittstellen:
Trainer, EventTrainer
public class GISTrainer extends AbstractEventTrainer
An implementation of Generalized Iterative Scaling (GIS).

The reference paper for this implementation was Adwait Ratnaparkhi's tech report at the University of Pennsylvania's Institute for Research in Cognitive Science, and is available at ftp://ftp.cis.upenn.edu/pub/ircs/tr/97-08.ps.Z.

The slack parameter used in the above implementation has been removed by default from the computation and a method for updating with Gaussian smoothing has been added per Investigating GIS and Smoothing for Maximum Entropy Taggers, Clark and Curran (2002). http://acl.ldc.upenn.edu/E/E03/E03-1071.pdf.

The slack parameter can be used by setting useSlackParameter to true. Gaussian smoothing can be used by setting useGaussianSmoothing to true.

A Prior can be used to train models which converge to the distribution which minimizes the relative entropy between the distribution specified by the empirical constraints of the training data and the specified prior. By default, the uniform distribution is used as the prior.

  • Felddetails

  • Konstruktordetails

    • GISTrainer

      public GISTrainer()
      Initializes a GISTrainer.

      Note:
      The resulting instance does not print progress messages about training to STDOUT.

  • Methodendetails

    • isSortAndMerge

      public boolean isSortAndMerge()
      Angegeben von:
      isSortAndMerge in Klasse AbstractEventTrainer

    • init

      public void init(TrainingParameters trainingParameters, Map<String,String> reportMap)
      Initializes a AbstractTrainer via TrainingParameters and a report map.
      Angegeben von:
      init in Schnittstelle Trainer
      Setzt außer Kraft:
      init in Klasse AbstractTrainer
      Parameter:
      trainingParameters - The TrainingParameters to use.
      reportMap - The Map instance used as report map.

    • doTrain

      public MaxentModel doTrain(DataIndexer indexer) throws IOException
      Angegeben von:
      doTrain in Klasse AbstractEventTrainer
      Löst aus:
      IOException

    • setSmoothing

      public void setSmoothing(boolean smooth)
      Sets whether this trainer will use smoothing while training the model.

      Note:
      This can improve model accuracy, though training will potentially take longer and use more memory. Model size will also be larger.

      Parameter:
      smooth - true if smoothing is desired, false if not.

    • setSmoothingObservation

      public void setSmoothingObservation(double timesSeen)
      Sets whether this trainer will use smoothing while training the model.

      Note:
      This can improve model accuracy, though training will potentially take longer and use more memory. Model size will also be larger.

      Parameter:
      timesSeen - The "number" of times we want the trainer to imagine it saw a feature that it actually didn't see

    • setGaussianSigma

      public void setGaussianSigma(double sigmaValue)
      Sets whether this trainer will use smoothing while training the model.

      Note:
      This can improve model accuracy, though training will potentially take longer and use more memory. Model size will also be larger.

      Parameter:
      sigmaValue - The Gaussian sigma value used for smoothing.

    • trainModel

      public GISModel trainModel(ObjectStream<Event> eventStream) throws IOException
      Trains a model using the GIS algorithm, assuming 100 iterations and no cutoff.
      Parameter:
      eventStream - The eventStream holding the data on which this model will be trained.
      Gibt zurück:
      A trained GISModel which can be used immediately or saved to disk using an GISModelWriter.
      Löst aus:
      IOException

    • trainModel

      public GISModel trainModel(ObjectStream<Event> eventStream, int iterations, int cutoff) throws IOException
      Trains a GIS model on the event in the specified event stream, using the specified number of iterations and the specified count cutoff.
      Parameter:
      eventStream - A stream of all events.
      iterations - The number of iterations to use for GIS.
      cutoff - The number of times a feature must occur to be included.
      Gibt zurück:
      A trained GISModel which can be used immediately or saved to disk using an GISModelWriter.
      Löst aus:
      IOException

    • trainModel

      public GISModel trainModel(int iterations, DataIndexer di)
      Trains a model using the GIS algorithm.
      Parameter:
      iterations - The number of GIS iterations to perform.
      di - The DataIndexer used to compress events in memory.
      Gibt zurück:
      A trained GISModel which can be used immediately or saved to disk using an GISModelWriter.
      Löst aus:
      IllegalArgumentException - Thrown if parameters were invalid.

    • trainModel

      public GISModel trainModel(int iterations, DataIndexer di, int threads)
      Trains a model using the GIS algorithm.
      Parameter:
      iterations - The number of GIS iterations to perform.
      di - The DataIndexer used to compress events in memory.
      threads - The number of thread to train with. Must be greater than 0.
      Gibt zurück:
      A trained GISModel which can be used immediately or saved to disk using an GISModelWriter.
      Löst aus:
      IllegalArgumentException - Thrown if parameters were invalid.

    • trainModel

      public GISModel trainModel(int iterations, DataIndexer di, Prior modelPrior, int threads)
      Trains a model using the GIS algorithm.
      Parameter:
      iterations - The number of GIS iterations to perform.
      di - The DataIndexer used to compress events in memory.
      modelPrior - The Prior distribution used to train this model.
      Gibt zurück:
      A trained GISModel which can be used immediately or saved to disk using an GISModelWriter.
      Löst aus:
      IllegalArgumentException - Thrown if parameters were invalid.