edu.stanford.nlp.ie.hmm
Class ContextTrainer

java.lang.Object
  |
  +--edu.stanford.nlp.ie.hmm.ContextTrainer

public class ContextTrainer
extends Object

Trains a context HMM on the contexts of the given target states, representing each target state as atomic. This utility trains on the entire given training data and builds an HMM with one background state and then transitions between it and the various target states in a potentially fully connected model. This will work well on data which is dense with states (bibliographic entries, addresses, ...), but doesn't do detailed state context modeling (in terms of prefix and suffix style background states) of the kind needed to do well on loosely structured natural language data.


Method Summary
static void main(String[] args)
          This is a command line utility that builds a serialized extractor for the context of a field.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Method Detail

main

public static void main(String[] args)
This is a command line utility that builds a serialized extractor for the context of a field. The usage is:
java ContextTrainer [-ct] trainingFile hmmFile targetField ...
or java ContextTrainer -cc trainingFile targetField ...
The first usage is the standard one for training a Context HMM. It saves the built hmm in a file hmmFile. The second usage just checks that isolateContext works by printing the first few documents, with target states replaced by a state token.

Parameters:
args - Command line arguments, as above.


Stanford NLP Group