edu.stanford.nlp.neural.rnn

Class TopNGramRecord

java.lang.Object

edu.stanford.nlp.neural.rnn.TopNGramRecord

public class TopNGramRecord
extends java.lang.Object

This class stores the best K ngrams for each class for a model. It does so by keeping priority queues of the best K trees seen, eliminating duplicates.
The interface is not too advanced at the moment. To use, first create it with the number of classes and the counts of the ngrams, and then add trees one at a time with countTree. The results can be extracted with toString(). There is no way to directly access the internal results.

Author:

John Bauer

Constructor Summary

Constructors

Constructor and Description
TopNGramRecord(int numClasses, int ngramCount, int maximumLength)

Method Summary

Modifier and Type	Method and Description
void	countTree(Tree tree) Adds the tree and all its subtrees to the appropriate PriorityQueues for each predicted class.
java.lang.String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

TopNGramRecord

public TopNGramRecord(int numClasses,
                      int ngramCount,
                      int maximumLength)

Method Detail

countTree

public void countTree(Tree tree)

Adds the tree and all its subtrees to the appropriate PriorityQueues for each predicted class.

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object