com.aliasi.lm
Class ScaleTrieReader

java.lang.Object
  com.aliasi.lm.ScaleTrieReader

All Implemented Interfaces:

TrieReader

public class ScaleTrieReader
extends Object

A ScaleTrieReader filters a contained trie reader by scaling all counts by a given multiple, removing all subtrees with zero root counts. Counts are rounded after multiplication using Math.round(double). Thus pruning will only occur if the scaling factor is less than 0.5, because 0.5 rounds to 1.

Since:

LingPipe2.3

Version:

2.3

Author:

Bob Carpenter

Constructor Summary

ScaleTrieReader(TrieReader reader, double scale)
Construct a scaling trie reader wrapping the specified reader and scaling counts with the specified value.

Method Summary

long	readCount() Returns the count of the next tree.
long	readSymbol() Returns the identifier of the symbol leading from the root of the current tree to the daughter subtree for the symbol, or -1 if there are no more subtrees for the current node.

Methods inherited from class java.lang.Object

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

Constructor Detail

ScaleTrieReader

public ScaleTrieReader(TrieReader reader,
                       double scale)
                throws IOException

Construct a scaling trie reader wrapping the specified reader and scaling counts with the specified value.

Parameters:

reader - Contained reader.

scale - Scaling factor.

Throws:

IllegalArgumentException - If the scale is not a positive, non-infinite value.

IOException

Method Detail

readCount

public long readCount()

Description copied from interface: TrieReader

Returns the count of the next tree.

Specified by:

readCount in interface TrieReader

Returns:

The count of the next tree.

readSymbol

public long readSymbol()
                throws IOException

Description copied from interface: TrieReader

Returns the identifier of the symbol leading from the root of the current tree to the daughter subtree for the symbol, or -1 if there are no more subtrees for the current node.

Specified by:

readSymbol in interface TrieReader

Returns:

The symbol leading to the next subtree.

Throws:

IOException - If there is an underlying I/O error.