private void UpdateCorrectionFromArc(TranslationResultBuilder builder, WordGraphArc arc, bool isPrefix,
int alignmentColsToAddCount)
{
for (int i = 0; i < arc.Words.Count; i++)
{
builder.AppendWord(arc.Words[i], arc.WordConfidences[i], !isPrefix && arc.IsUnknown);
}
WordAlignmentMatrix alignment = arc.Alignment;
if (alignmentColsToAddCount > 0)
{
var newAlignment = new WordAlignmentMatrix(alignment.RowCount,
alignment.ColumnCount + alignmentColsToAddCount);
for (int j = 0; j < alignment.ColumnCount; j++)
{
for (int i = 0; i < alignment.RowCount; i++)
{
newAlignment[i, alignmentColsToAddCount + j] = alignment[i, j];
}
}
alignment = newAlignment;
}
builder.MarkPhrase(arc.SourceSegmentRange, alignment);
}
public WordAlignmentMatrix GetBestAlignment(IReadOnlyList <string> sourceSegment, IReadOnlyList <string> targetSegment,
WordAlignmentMatrix hintMatrix = null)
{
var paa = new PairwiseAlignmentAlgorithm <IReadOnlyList <string>, int>(_scorer, sourceSegment, targetSegment, GetWordIndices)
{
Mode = AlignmentMode.Global,
ExpansionCompressionEnabled = true,
TranspositionEnabled = true
};
paa.Compute();
Alignment <IReadOnlyList <string>, int> alignment = paa.GetAlignments().First();
var waMatrix = new WordAlignmentMatrix(sourceSegment.Count, targetSegment.Count);
for (int c = 0; c < alignment.ColumnCount; c++)
{
foreach (int i in alignment[0, c])
{
foreach (int j in alignment[1, c])
{
waMatrix[i, j] = AlignmentType.Aligned;
}
}
}
return(waMatrix);
}
public void SymmetrizeWith(WordAlignmentMatrix other)
{
if (RowCount != other.RowCount || ColumnCount != other.ColumnCount)
{
throw new ArgumentException("The matrices are not the same size.", nameof(other));
}
WordAlignmentMatrix aux = Clone();
IntersectWith(other);
WordAlignmentMatrix prev = null;
while (!ValueEquals(prev))
{
prev = Clone();
for (int i = 0; i < RowCount; i++)
{
for (int j = 0; j < ColumnCount; j++)
{
if ((other._matrix[i, j] == AlignmentType.Aligned || aux._matrix[i, j] == AlignmentType.Aligned) && _matrix[i, j] == AlignmentType.NotAligned)
{
if (IsColumnAligned(j) == AlignmentType.NotAligned && IsRowAligned(i) == AlignmentType.NotAligned)
{
_matrix[i, j] = AlignmentType.Aligned;
}
else if (IsNeighborAligned(i, j))
{
_matrix[i, j] = AlignmentType.Aligned;
}
}
}
}
}
}
public TranslationResult(IEnumerable <string> sourceSegment, IEnumerable <string> targetSegment,
IEnumerable <double> confidences, IEnumerable <TranslationSources> sources, WordAlignmentMatrix alignment,
IEnumerable <Phrase> phrases)
{
SourceSegment = sourceSegment.ToArray();
TargetSegment = targetSegment.ToArray();
WordConfidences = confidences.ToArray();
if (WordConfidences.Count != TargetSegment.Count)
{
throw new ArgumentException("The confidences must be the same length as the target segment.",
nameof(confidences));
}
WordSources = sources.ToArray();
if (WordSources.Count != TargetSegment.Count)
{
throw new ArgumentException("The sources must be the same length as the target segment.",
nameof(sources));
}
Alignment = alignment;
if (Alignment.RowCount != SourceSegment.Count)
{
throw new ArgumentException(
"The alignment source length must be the same length as the source segment.", nameof(alignment));
}
if (Alignment.ColumnCount != TargetSegment.Count)
{
throw new ArgumentException(
"The alignment target length must be the same length as the target segment.", nameof(alignment));
}
Phrases = phrases.ToArray();
}
public static WordAlignmentMatrix CreateAlignmentMatrix(this ParallelTextSegment segment, bool isUnknown = true)
{
if (segment.AlignedWordPairs == null)
{
return(null);
}
var matrix = new WordAlignmentMatrix(segment.SourceSegment.Count, segment.TargetSegment.Count,
isUnknown ? AlignmentType.Unknown : AlignmentType.NotAligned);
foreach (AlignedWordPair wordPair in segment.AlignedWordPairs)
{
matrix[wordPair.SourceIndex, wordPair.TargetIndex] = AlignmentType.Aligned;
if (isUnknown)
{
for (int i = 0; i < segment.SourceSegment.Count; i++)
{
if (matrix[i, wordPair.TargetIndex] == AlignmentType.Unknown)
{
matrix[i, wordPair.TargetIndex] = AlignmentType.NotAligned;
}
}
for (int j = 0; j < segment.TargetSegment.Count; j++)
{
if (matrix[wordPair.SourceIndex, j] == AlignmentType.Unknown)
{
matrix[wordPair.SourceIndex, j] = AlignmentType.NotAligned;
}
}
}
}
return(matrix);
}
public WordAlignmentMatrix Approve()
{
CheckDisposed();
WordAlignmentMatrix matrix = _engine.GetHintMatrix(SourceSegment, Prefix, _ruleResult);
_engine.SmtEngine.TrainSegment(SourceSegment, Prefix, matrix);
return(matrix);
}
public IEnumerable <TranslationResult> Translate(int n, IReadOnlyList <string> segment)
{
IEnumerable <IEnumerable <WordAnalysis> > sourceAnalyses = segment
.Select(word => _sourceAnalyzer.AnalyzeWord(word));
foreach (TransferResult transferResult in _transferer.Transfer(sourceAnalyses).Take(n))
{
IReadOnlyList <WordAnalysis> targetAnalyses = transferResult.TargetAnalyses;
WordAlignmentMatrix waMatrix = transferResult.WordAlignmentMatrix;
var translation = new List <string>();
var confidences = new List <double>();
var sources = new List <TranslationSources>();
var alignment = new WordAlignmentMatrix(segment.Count, targetAnalyses.Count);
double confidence = double.MaxValue;
for (int j = 0; j < targetAnalyses.Count; j++)
{
int[] sourceIndices = Enumerable.Range(0, waMatrix.RowCount)
.Where(i => waMatrix[i, j]).ToArray();
string targetWord = targetAnalyses[j].IsEmpty
? null
: _targetGenerator.GenerateWords(targetAnalyses[j]).FirstOrDefault();
double wordConfidence = 1.0;
TranslationSources source = TranslationSources.Transfer;
if (targetWord == null)
{
if (sourceIndices.Length > 0)
{
int i = sourceIndices[0];
targetWord = segment[i];
wordConfidence = 0;
source = TranslationSources.None;
alignment[i, j] = true;
}
}
else
{
foreach (int i in sourceIndices)
{
alignment[i, j] = true;
}
}
if (targetWord != null)
{
translation.Add(targetWord);
confidences.Add(wordConfidence);
sources.Add(source);
confidence = Math.Min(confidence, wordConfidence);
}
}
yield return(new TranslationResult(segment, translation, confidences, sources, alignment,
new[] { new Phrase(Range <int> .Create(0, segment.Count), translation.Count, confidence) }));
}
}
public void AddSegmentPair(IReadOnlyList <string> sourceSegment, IReadOnlyList <string> targetSegment,
WordAlignmentMatrix hintMatrix = null)
{
_directWordAlignmentModel.AddSegmentPair(sourceSegment, targetSegment, hintMatrix);
WordAlignmentMatrix invertedHintMatrix = hintMatrix?.Clone();
invertedHintMatrix?.Transpose();
_inverseWordAlignmentModel.AddSegmentPair(targetSegment, sourceSegment, invertedHintMatrix);
}
public WordAlignmentMatrix GetBestAlignment(IReadOnlyList <string> sourceSegment,
IReadOnlyList <string> targetSegment)
{
WordAlignmentMatrix matrix = _srcTrgAligner.GetBestAlignment(sourceSegment, targetSegment);
WordAlignmentMatrix invMatrix = _trgSrcAligner.GetBestAlignment(targetSegment, sourceSegment);
invMatrix.Transpose();
matrix.SymmetrizeWith(invMatrix);
return(matrix);
}
public static string GetGizaFormatString(this ISegmentAligner aligner, ParallelTextSegment segment,
Func <string, string> sourcePreprocessor = null, Func <string, string> targetPreprocessor = null)
{
IReadOnlyList <string> sourceSegment = segment.SourceSegment.Preprocess(sourcePreprocessor);
IReadOnlyList <string> targetSegment = segment.TargetSegment.Preprocess(targetPreprocessor);
WordAlignmentMatrix alignment = aligner.GetBestAlignment(sourceSegment, targetSegment,
segment.CreateAlignmentMatrix());
return(alignment.ToGizaFormat(sourceSegment, targetSegment));
}
private WordAlignmentMatrix(WordAlignmentMatrix other)
{
_matrix = new AlignmentType[other.RowCount, other.ColumnCount];
for (int i = 0; i < RowCount; i++)
{
for (int j = 0; j < ColumnCount; j++)
{
_matrix[i, j] = other._matrix[i, j];
}
}
}
public WordAlignmentMatrix TrainSegment(IReadOnlyList <string> sourceSegment,
IReadOnlyList <string> targetSegment)
{
CheckDisposed();
TranslationResult ruleResult = RuleEngine?.Translate(sourceSegment);
WordAlignmentMatrix matrix = GetHintMatrix(sourceSegment, targetSegment, ruleResult);
SmtEngine.TrainSegment(sourceSegment, targetSegment, matrix);
return(matrix);
}
public static string RecaseTargetWord(this WordAlignmentMatrix alignment, IReadOnlyList <string> sourceSegment,
int sourceStartIndex, IReadOnlyList <string> targetSegment, int targetIndex)
{
string targetWord = targetSegment[targetIndex];
if (alignment.GetColumnAlignedIndices(targetIndex)
.Any(i => sourceSegment[sourceStartIndex + i].IsTitleCase()))
{
return(targetWord.ToTitleCase());
}
return(targetWord);
}
public static WordAlignmentMatrix GetBestAlignment(this ISegmentAligner aligner,
IReadOnlyList <string> sourceSegment, IReadOnlyList <string> targetSegment,
WordAlignmentMatrix knownAlignment)
{
WordAlignmentMatrix estimatedAlignment = aligner.GetBestAlignment(sourceSegment, targetSegment);
WordAlignmentMatrix alignment = estimatedAlignment;
if (knownAlignment != null)
{
alignment = knownAlignment.Clone();
alignment.PrioritySymmetrizeWith(estimatedAlignment);
}
return(alignment);
}
public TranslationResult ToResult(IReadOnlyList <string> sourceSegment, int prefixCount = 0)
{
double[] confidences = _confidences.ToArray();
var sources = new TranslationSources[Words.Count];
var alignment = new WordAlignmentMatrix(sourceSegment.Count, Words.Count);
var phrases = new List <Phrase>();
int trgPhraseStartIndex = 0;
foreach (PhraseInfo phraseInfo in _phrases)
{
double confidence = double.MaxValue;
for (int j = trgPhraseStartIndex; j < phraseInfo.TargetCut; j++)
{
for (int i = phraseInfo.SourceSegmentRange.Start; i < phraseInfo.SourceSegmentRange.End; i++)
{
AlignmentType alignmentType = phraseInfo.Alignment[i - phraseInfo.SourceSegmentRange.Start,
j - trgPhraseStartIndex];
if (alignmentType == AlignmentType.Aligned)
{
alignment[i, j] = AlignmentType.Aligned;
}
}
if (j < prefixCount)
{
sources[j] = TranslationSources.Prefix;
if (_uncorrectedPrefixWords.Contains(j))
{
sources[j] |= TranslationSources.Smt;
}
}
else if (_unknownWords.Contains(j))
{
sources[j] = TranslationSources.None;
}
else
{
sources[j] = TranslationSources.Smt;
}
confidence = Math.Min(confidence, Confidences[j]);
}
phrases.Add(new Phrase(phraseInfo.SourceSegmentRange, phraseInfo.TargetCut, confidence));
trgPhraseStartIndex = phraseInfo.TargetCut;
}
return(new TranslationResult(sourceSegment, Words, confidences, sources, alignment, phrases));
}
internal WordAlignmentMatrix GetHintMatrix(IReadOnlyList <string> sourceSegment,
IReadOnlyList <string> targetSegment, TranslationResult ruleResult)
{
TranslationResult smtResult = SmtEngine.GetBestPhraseAlignment(sourceSegment, targetSegment);
TranslationResult hybridResult = ruleResult == null ? smtResult : smtResult.Merge(targetSegment.Count,
RuleEngineThreshold, ruleResult);
var matrix = new WordAlignmentMatrix(sourceSegment.Count, targetSegment.Count, AlignmentType.Unknown);
var iAligned = new HashSet <int>();
for (int j = 0; j < targetSegment.Count; j++)
{
bool jAligned = false;
if (j < hybridResult.WordSources.Count &&
(hybridResult.WordSources[j] & TranslationSources.Transfer) != 0)
{
foreach (int i in hybridResult.Alignment.GetColumnAlignedIndices(j))
{
matrix[i, j] = AlignmentType.Aligned;
iAligned.Add(i);
jAligned = true;
}
}
if (jAligned)
{
for (int i = 0; i < sourceSegment.Count; i++)
{
if (matrix[i, j] == AlignmentType.Unknown)
{
matrix[i, j] = AlignmentType.NotAligned;
}
}
}
}
foreach (int i in iAligned)
{
for (int j = 0; j < targetSegment.Count; j++)
{
if (matrix[i, j] == AlignmentType.Unknown)
{
matrix[i, j] = AlignmentType.NotAligned;
}
}
}
return(matrix);
}
public static string GetAlignmentString(this IWordAlignmentModel model, ParallelTextSegment segment,
bool includeProbs, Func <string, string> sourcePreprocessor = null,
Func <string, string> targetPreprocessor = null)
{
IReadOnlyList <string> sourceSegment = segment.SourceSegment.Preprocess(sourcePreprocessor);
IReadOnlyList <string> targetSegment = segment.TargetSegment.Preprocess(targetPreprocessor);
WordAlignmentMatrix alignment = model.GetBestAlignment(sourceSegment, targetSegment,
segment.CreateAlignmentMatrix());
if (includeProbs)
{
return(alignment.ToString(model, sourceSegment, targetSegment));
}
return(alignment.ToString());
}
private static TranslationResult CreateResult(int sourceLen, int prefixLen, string target,
params double[] confidences)
{
string[] targetArray = target.Split();
var targetConfidences = new double[targetArray.Length];
var targetSources = new TranslationSources[targetArray.Length];
var alignment = new WordAlignmentMatrix(sourceLen, targetArray.Length);
int i = 0, j = 0;
double phraseConfidence = double.MaxValue;
foreach (double confidence in confidences)
{
if (j < prefixLen)
{
targetSources[j] = TranslationSources.Prefix;
}
if (confidence >= 0)
{
alignment[i, j] = true;
targetConfidences[j] = confidence;
if (confidence > 0)
{
targetSources[j] |= TranslationSources.Smt;
}
i++;
j++;
}
else if (targetArray.Length > sourceLen)
{
targetConfidences[j] = confidence;
j++;
}
else if (targetArray.Length < sourceLen)
{
i++;
}
else
{
throw new ArgumentException("A confidence was incorrectly set below 0.", nameof(confidences));
}
phraseConfidence = Math.Min(phraseConfidence, confidence);
}
return(new TranslationResult(Enumerable.Range(0, sourceLen).Select(index => index.ToString()), targetArray,
targetConfidences, targetSources, alignment,
new[] { new Phrase(Range <int> .Create(0, sourceLen), targetArray.Length, phraseConfidence) }));
}
public static WordAlignmentMatrix CreateAlignmentMatrix(this ParallelTextSegment segment)
{
if (segment.AlignedWordPairs == null)
{
return(null);
}
var matrix = new WordAlignmentMatrix(segment.SourceSegment.Count, segment.TargetSegment.Count);
foreach (AlignedWordPair wordPair in segment.AlignedWordPairs)
{
matrix[wordPair.SourceIndex, wordPair.TargetIndex] = true;
}
return(matrix);
}
public IEnumerable <TransferResult> Transfer(IEnumerable <IEnumerable <WordAnalysis> > sourceAnalyses)
{
var targetAnalyses = new List <WordAnalysis>();
foreach (IEnumerable <WordAnalysis> sourceAnalysisOptions in sourceAnalyses)
{
bool found = false;
foreach (WordAnalysis sourceAnalysisOption in sourceAnalysisOptions)
{
var targetMorphemes = new List <IMorpheme>();
foreach (IMorpheme sourceMorpheme in sourceAnalysisOption.Morphemes)
{
IMorpheme targetMorpheme;
if (!_morphemeMapper.TryGetTargetMorpheme(sourceMorpheme, out targetMorpheme))
{
break;
}
targetMorphemes.Add(targetMorpheme);
}
if (targetMorphemes.Count == sourceAnalysisOption.Morphemes.Count)
{
targetAnalyses.Add(new WordAnalysis(targetMorphemes, sourceAnalysisOption.RootMorphemeIndex, sourceAnalysisOption.Category));
found = true;
break;
}
}
if (!found)
{
targetAnalyses.Add(new WordAnalysis());
}
}
var waMatrix = new WordAlignmentMatrix(targetAnalyses.Count, targetAnalyses.Count);
for (int j = 0; j < targetAnalyses.Count; j++)
{
waMatrix[j, j] = true;
}
var result = new TransferResult(targetAnalyses, waMatrix);
return(result.ToEnumerable());
}
public void IntersectWith(WordAlignmentMatrix other)
{
if (RowCount != other.RowCount || ColumnCount != other.ColumnCount)
{
throw new ArgumentException("The matrices are not the same size.", nameof(other));
}
for (int i = 0; i < RowCount; i++)
{
for (int j = 0; j < ColumnCount; j++)
{
if (!(_matrix[i, j] == AlignmentType.Aligned && other._matrix[i, j] == AlignmentType.Aligned))
{
_matrix[i, j] = AlignmentType.NotAligned;
}
}
}
}
public WordAlignmentMatrix GetBestAlignment(IReadOnlyList <string> sourceSegment, IReadOnlyList <string> targetSegment,
WordAlignmentMatrix hintMatrix = null)
{
WordAlignmentMatrix matrix = _srcTrgAligner.GetBestAlignment(sourceSegment, targetSegment, hintMatrix);
WordAlignmentMatrix invHintMatrix = null;
if (hintMatrix != null)
{
invHintMatrix = hintMatrix.Clone();
invHintMatrix.Transpose();
}
WordAlignmentMatrix invMatrix = _trgSrcAligner.GetBestAlignment(targetSegment, sourceSegment, invHintMatrix);
invMatrix.Transpose();
matrix.SymmetrizeWith(invMatrix);
return(matrix);
}
public WordGraphArc(int prevState, int nextState, double score, IEnumerable <string> words,
WordAlignmentMatrix alignment, Range <int> sourceSegmentRange, bool isUnknown,
IEnumerable <double> wordConfidences = null)
{
PrevState = prevState;
NextState = nextState;
Score = score;
Words = words.ToArray();
Alignment = alignment;
SourceSegmentRange = sourceSegmentRange;
IsUnknown = isUnknown;
if (wordConfidences == null)
{
WordConfidences = Enumerable.Repeat(-1.0, Words.Count).ToList();
}
else
{
WordConfidences = wordConfidences.ToList();
}
}
private void ResizeAlignment(int phraseIndex, List <int> colsToCopy)
{
WordAlignmentMatrix curAlignment = _phrases[phraseIndex].Alignment;
if (colsToCopy.Count == curAlignment.ColumnCount)
{
return;
}
var newAlignment = new WordAlignmentMatrix(curAlignment.RowCount, colsToCopy.Count);
for (int j = 0; j < newAlignment.ColumnCount; j++)
{
if (colsToCopy[j] != -1)
{
for (int i = 0; i < newAlignment.RowCount; i++)
{
newAlignment[i, j] = curAlignment[i, colsToCopy[j]];
}
}
}
_phrases[phraseIndex].Alignment = newAlignment;
}
public bool ValueEquals(WordAlignmentMatrix other)
{
if (other == null)
{
return(false);
}
if (RowCount != other.RowCount || ColumnCount != other.ColumnCount)
{
return(false);
}
for (int i = 0; i < RowCount; i++)
{
for (int j = 0; j < ColumnCount; j++)
{
if (_matrix[i, j] != other._matrix[i, j])
{
return(false);
}
}
}
return(true);
}
public WordAlignmentMatrix GetBestAlignment(IReadOnlyList <string> sourceSegment,
IReadOnlyList <string> targetSegment, WordAlignmentMatrix hintMatrix = null)
{
var paa = new PairwiseAlignmentAlgorithm <IReadOnlyList <string>, int>(_scorer, sourceSegment, targetSegment,
GetWordIndices)
{
Mode = AlignmentMode.Global,
ExpansionCompressionEnabled = true,
TranspositionEnabled = true
};
paa.Compute();
Alignment <IReadOnlyList <string>, int> alignment = paa.GetAlignments().First();
var waMatrix = new WordAlignmentMatrix(sourceSegment.Count, targetSegment.Count);
for (int c = 0; c < alignment.ColumnCount; c++)
{
foreach (int j in alignment[1, c])
{
double bestScore;
int minIndex, maxIndex;
if (alignment[0, c].IsNull)
{
double prob = _getTranslationProb(null, targetSegment[j]);
bestScore = ComputeAlignmentScore(prob, 0);
int tc = c - 1;
while (tc >= 0 && alignment[0, tc].IsNull)
{
tc--;
}
int i = tc == -1 ? 0 : alignment[0, tc].Last;
minIndex = i;
maxIndex = i + 1;
}
else
{
double prob = alignment[0, c]
.Average(i => _getTranslationProb(sourceSegment[i], targetSegment[j]));
bestScore = ComputeAlignmentScore(prob, 0);
minIndex = alignment[0, c].First - 1;
maxIndex = alignment[0, c].Last + 1;
}
int bestIndex = -1;
for (int i = minIndex; i >= Math.Max(0, minIndex - _maxDistance); i--)
{
double prob = _getTranslationProb(sourceSegment[i], targetSegment[j]);
double distanceScore = ComputeDistanceScore(i, minIndex + 1, sourceSegment.Count);
double score = ComputeAlignmentScore(prob, distanceScore);
if (score > bestScore)
{
bestScore = score;
bestIndex = i;
}
}
for (int i = maxIndex; i < Math.Min(sourceSegment.Count, maxIndex + _maxDistance); i++)
{
double prob = _getTranslationProb(sourceSegment[i], targetSegment[j]);
double distanceScore = ComputeDistanceScore(i, maxIndex - 1, sourceSegment.Count);
double score = ComputeAlignmentScore(prob, distanceScore);
if (score > bestScore)
{
bestScore = score;
bestIndex = i;
}
}
if (bestIndex == -1)
{
if (!alignment[0, c].IsNull)
{
waMatrix[minIndex + 1, j] = AlignmentType.Aligned;
waMatrix[maxIndex - 1, j] = AlignmentType.Aligned;
}
}
else
{
waMatrix[bestIndex, j] = AlignmentType.Aligned;
}
}
}
return(waMatrix);
}
public PhraseInfo(Range <int> sourceSegmentRange, int targetCut, WordAlignmentMatrix alignment)
{
SourceSegmentRange = sourceSegmentRange;
TargetCut = targetCut;
Alignment = alignment;
}
public WordAlignmentMatrix GetBestAlignment(IReadOnlyList <string> sourceSegment,
IReadOnlyList <string> targetSegment, WordAlignmentMatrix hintMatrix = null)
{
CheckDisposed();
return(_aligner.GetBestAlignment(sourceSegment, targetSegment, hintMatrix));
}
public void MarkPhrase(Range <int> sourceSegmentRange, WordAlignmentMatrix alignment)
{
_phrases.Add(new PhraseInfo(sourceSegmentRange, _words.Count, alignment));
}
public TranslationResult Merge(int prefixCount, double threshold, TranslationResult otherResult)
{
var mergedTargetSegment = new List <string>();
var mergedConfidences = new List <double>();
var mergedSources = new List <TranslationSources>();
var mergedAlignment = new HashSet <Tuple <int, int> >();
for (int j = 0; j < TargetSegment.Count; j++)
{
int[] sourceIndices = Alignment.GetColumnAlignedIndices(j).ToArray();
if (sourceIndices.Length == 0)
{
// target word doesn't align with anything
mergedTargetSegment.Add(TargetSegment[j]);
mergedConfidences.Add(WordConfidences[j]);
mergedSources.Add(WordSources[j]);
}
else
{
// target word aligns with some source words
if (j < prefixCount || WordConfidences[j] >= threshold)
{
// use target word of this result
mergedTargetSegment.Add(TargetSegment[j]);
mergedConfidences.Add(WordConfidences[j]);
TranslationSources sources = WordSources[j];
foreach (int i in sourceIndices)
{
// combine sources for any words that both this result
// and the other result translated the same
foreach (int jOther in otherResult.Alignment.GetRowAlignedIndices(i))
{
TranslationSources otherSources = otherResult.WordSources[jOther];
if (otherSources != TranslationSources.None &&
otherResult.TargetSegment[jOther] == TargetSegment[j])
{
sources |= otherSources;
}
}
mergedAlignment.Add(Tuple.Create(i, mergedTargetSegment.Count - 1));
}
mergedSources.Add(sources);
}
else
{
// use target words of other result
bool found = false;
foreach (int i in sourceIndices)
{
foreach (int jOther in otherResult.Alignment.GetRowAlignedIndices(i))
{
// look for any translated words from other result
TranslationSources otherSources = otherResult.WordSources[jOther];
if (otherSources != TranslationSources.None)
{
mergedTargetSegment.Add(otherResult.TargetSegment[jOther]);
mergedConfidences.Add(otherResult.WordConfidences[jOther]);
mergedSources.Add(otherSources);
mergedAlignment.Add(Tuple.Create(i, mergedTargetSegment.Count - 1));
found = true;
}
}
}
if (!found)
{
// the other result had no translated words, so just use this result's target word
mergedTargetSegment.Add(TargetSegment[j]);
mergedConfidences.Add(WordConfidences[j]);
mergedSources.Add(WordSources[j]);
foreach (int i in sourceIndices)
{
mergedAlignment.Add(Tuple.Create(i, mergedTargetSegment.Count - 1));
}
}
}
}
}
var alignment = new WordAlignmentMatrix(SourceSegment.Count, mergedTargetSegment.Count);
foreach (Tuple <int, int> t in mergedAlignment)
{
alignment[t.Item1, t.Item2] = true;
}
return(new TranslationResult(SourceSegment, mergedTargetSegment, mergedConfidences, mergedSources,
alignment, Phrases));
}
