public void TrainSegment_TranslationCorrect()
{
using (var smtModel = new ThotSmtModel(TestHelpers.ToyCorpusConfigFileName))
using (ISmtEngine engine = smtModel.CreateEngine())
{
TranslationResult result = engine.Translate("esto es una prueba .".Split());
Assert.That(result.TargetSegment, Is.EqualTo("esto is a prueba .".Split()));
engine.TrainSegment("esto es una prueba .".Split(), "this is a test .".Split());
result = engine.Translate("esto es una prueba .".Split());
Assert.That(result.TargetSegment, Is.EqualTo("this is a test .".Split()));
}
}
public void TrainSegment_AlignmentSpecified_TranslationCorrect()
{
using (var smtModel = new ThotSmtModel(TestHelpers.ToyCorpusConfigFileName))
using (ISmtEngine engine = smtModel.CreateEngine())
{
TranslationResult result = engine.Translate("maria no dio una bofetada a la bruja verde .".Split());
Assert.That(result.TargetSegment, Is.EqualTo("maria no dio a bofetada to bruja verde .".Split()));
var matrix = new WordAlignmentMatrix(10, 7, AlignmentType.Unknown);
SetAligned(matrix, 1, 1);
SetAligned(matrix, 2, 2);
SetAligned(matrix, 3, 2);
SetAligned(matrix, 4, 2);
SetSourceNotAligned(matrix, 5);
SetAligned(matrix, 8, 4);
engine.TrainSegment("maria no dio una bofetada a la bruja verde .".Split(), "mary didn't slap the green witch .".Split(), matrix);
result = engine.Translate("maria es una bruja .".Split());
Assert.That(result.TargetSegment, Is.EqualTo("mary is a witch .".Split()));
}
}
private void TrainTuneCorpus(string trainTMPrefix, string trainLMPrefix,
IReadOnlyList <IReadOnlyList <string> > tuneSourceCorpus,
IReadOnlyList <IReadOnlyList <string> > tuneTargetCorpus, ThotTrainProgressReporter reporter)
{
reporter.Step("Finalizing", TrainingStepCount - 1);
if (tuneSourceCorpus.Count == 0)
{
return;
}
ThotSmtParameters parameters = Parameters.Clone();
parameters.TranslationModelFileNamePrefix = trainTMPrefix;
parameters.LanguageModelFileNamePrefix = trainLMPrefix;
using (var smtModel = new ThotSmtModel(parameters))
using (ISmtEngine engine = smtModel.CreateEngine())
{
for (int i = 0; i < tuneSourceCorpus.Count; i++)
{
engine.TrainSegment(tuneSourceCorpus[i], tuneTargetCorpus[i]);
}
}
}
private void TrainTuneCorpus(string trainTMPrefix, string trainLMPrefix,
IReadOnlyList <IReadOnlyList <string> > tuneSourceCorpus,
IReadOnlyList <IReadOnlyList <string> > tuneTargetCorpus, IProgress <ProgressStatus> progress)
{
if (tuneSourceCorpus.Count == 0)
{
return;
}
ThotSmtParameters parameters = Parameters.Clone();
parameters.TranslationModelFileNamePrefix = trainTMPrefix;
parameters.LanguageModelFileNamePrefix = trainLMPrefix;
using (var smtModel = new ThotSmtModel(parameters))
using (ISmtEngine engine = smtModel.CreateEngine())
{
for (int i = 0; i < tuneSourceCorpus.Count; i++)
{
progress.Report(new ProgressStatus(i, tuneSourceCorpus.Count));
engine.TrainSegment(tuneSourceCorpus[i], tuneTargetCorpus[i]);
}
progress.Report(new ProgressStatus(1.0));
}
}