MALLET中的分类器

1.concept:Classifiers map from instances to distributions over a fixed set of classes

2.methods:MaxEnt,
Naïve Bayes,
Decision Trees...

3.

public void testInstance() { ArrayList<Pipe> pipeList = new ArrayList<Pipe>(); pipeList.add(new Target2Label()); pipeList.add(new CharSequence2TokenSequence()); pipeList.add(new TokenSequence2FeatureSequence()); pipeList.add(new FeatureSequence2FeatureVector()); InstanceList instances = new InstanceList(new SerialPipes(pipeList)); ClassifierTrainer trainer = new MaxEntTrainer(); Classifier classifier = trainer.train(instances); } 

4.optimize:

package cc.mallet.optimize; public interface Optimizable{} public interface Optimizer{} 

5.Create random test/train splits

InstanceList[] instanceLists = instances.split(new Randoms(), new double[] { 0.9, .1, .0 }); 

means:90% for training,10% for testing,0% for validation

6.The
Trial Stores the results of classifying a collection of Instances, and provides many methods for evaluating the results.