public static WordAlignmentMatrix GetBestAlignment(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);
return(aligner.GetBestAlignment(sourceSegment, targetSegment, segment.CreateAlignmentMatrix()));
}
public ThotSmtEngine(ThotSmtModel smtModel)
{
_smtModel = smtModel;
_sessions = new HashSet <ThotInteractiveTranslationSession>();
LoadHandle();
_segmentAligner = new FuzzyEditDistanceSegmentAligner(GetTranslationProbability);
_confidenceEstimator = new Ibm1WordConfidenceEstimator(GetTranslationProbability);
//_confidenceEstimator = new WppWordConfidenceEstimator(this);
ErrorCorrectionModel = new ErrorCorrectionModel();
}
public static string GetGizaFormatString(this ISegmentAligner aligner, ParallelTextSegment segment,
Func <string, string> sourcePreprocessor = null, Func <string, string> targetPreprocessor = null,
bool isUnknown = true)
{
IReadOnlyList <string> sourceSegment = segment.SourceSegment.Preprocess(sourcePreprocessor);
IReadOnlyList <string> targetSegment = segment.TargetSegment.Preprocess(targetPreprocessor);
WordAlignmentMatrix alignment = aligner.GetBestAlignment(sourceSegment, targetSegment,
segment.CreateAlignmentMatrix(isUnknown));
return(alignment.ToGizaFormat(sourceSegment, targetSegment));
}
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 SymmetrizedSegmentAligner(ISegmentAligner srcTrgAligner, ISegmentAligner trgSrcAligner)
{
_srcTrgAligner = srcTrgAligner;
_trgSrcAligner = trgSrcAligner;
}
