edu.washington.cs.knowitall.extractor.conf
Class ReVerbClassifierTrainer
java.lang.Object
edu.washington.cs.knowitall.extractor.conf.ReVerbClassifierTrainer
public class ReVerbClassifierTrainer
- extends java.lang.Object
Used to train the ReVerb confidence function using the features described
by ReVerbFeatures
. Given a set of LabeledBinaryExtraction
instances, this class featurizes them and trains a logistic regression classifier
using Weka's Logistic
class.
This class can be called from the command-line to train a classifier and save the
resulting model to a file.
- Author:
- afader
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
ReVerbClassifierTrainer
public ReVerbClassifierTrainer(java.lang.Iterable<LabeledBinaryExtraction> examples)
throws java.lang.Exception
- Constructs and trains a new Logistic classifier using the given examples.
- Parameters:
examples
-
- Throws:
java.lang.Exception
getDataSet
public WekaDataSet<ChunkedBinaryExtraction> getDataSet()
- Returns:
- the data set used to train the classifier
getClassifier
public weka.classifiers.functions.Logistic getClassifier()
- Returns:
- the trained classifier.
main
public static void main(java.lang.String[] args)
throws java.lang.Exception
- Trains a logistic regression classifier using the examples in the given file,
and saves the model to disk. The examples must be in the format described in
LabeledBinaryExtractionReader
.
An optional third parameter can be passed that writes the training data in
Weka's ARFF file format to disk.
- Parameters:
args
-
- Throws:
java.lang.Exception