edu.umn.cs.nlp.mt
Class AlignmentMatrix
java.lang.Object
edu.umn.cs.nlp.mt.AlignmentMatrix
public class AlignmentMatrix
- extends Object
Represents a word alignment matrix for an aligned sentence pair in a machine translation system.
Initial phrase pairs and hierarchical rules may be extracted from an alignment matrix,
based on the constraints described in (Chiang, 2005).
- Version:
- $LastChangedDate: 2007-09-11 08:39:02 -0500 (Tue, 11 Sep 2007) $
- Author:
- Lane Schwartz
- See Also:
- "A Hierarchical Phrase-Based Model for Statistical Machine Translation" by David Chiang (ACL, 2005)
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
MAX_SENTENCE_LENGTH
public static final int MAX_SENTENCE_LENGTH
- See Also:
- Constant Field Values
DELIMITER
public static final String DELIMITER
- See Also:
- Constant Field Values
AlignmentMatrix
public AlignmentMatrix(String sourceLine,
String targetLine,
String alignmentLine,
LexicalTranslationProbabilityDistribution lexProbs)
main
public static void main(String[] args)
getBasicPhrasePairs
protected List<PhrasePair> getBasicPhrasePairs()
getCompositePhrasePairs
protected List<PhrasePair> getCompositePhrasePairs()
extractBasicPhrasePairs
public void extractBasicPhrasePairs()
extractCompositePhrasePairs
public void extractCompositePhrasePairs()
getPhrasePairsAsString
public String getPhrasePairsAsString()
phrasePairToString
public String phrasePairToString(PhrasePair p)
- Convert a
PhrasePair
to a String
.
The returned string will have the following format:
source_phrase ::: target_phrase ::: source_given_target_lex_prob ::: target_given_source_lex_prob ::: weight
- Parameters:
p
-
- Returns:
getNumberExtractedFrom
protected double getNumberExtractedFrom(PhrasePair p)
phrasePairToString
public String phrasePairToString(PhrasePair p,
PhrasePair nonterminal)
phrasePairToString
public String phrasePairToString(PhrasePair p,
PhrasePair nt1,
PhrasePair nt2)