static void Main(string[] args)
{
TranslationTrainingData trainingData = TranslationTrainingData.AlienLanguage;
IBMModel1Dictionary dictionary = IBMModel1Dictionary.Train(trainingData, 20);
foreach (string f in trainingData.FWords)
{
KeyValuePair <string, float>[] translations = dictionary.Table[f]
.Select((pair, index) => { return(pair); })
.ToArray();
TranslationComparer comparer = new TranslationComparer();
Array.Sort(translations, comparer);
Array.Reverse(translations);
IEnumerable <string> bestTranslations = translations
.Take(3)
.Select((pair, index) => $"{pair.Key}({pair.Value:0.00})");
Console.WriteLine("(f = {0}): {1}",
f,
string.Join(", ", bestTranslations));
}
}
public static IBMModel1Dictionary Train(
TranslationTrainingData trainingData,
int iterations = 10,
float defaultTableValue = 1.0f)
{
IBMModel1Dictionary dictionary = new IBMModel1Dictionary();
// Initialize dictionary
foreach (string f in trainingData.FWords)
{
foreach (string e in trainingData.EWords)
{
dictionary.Table.InitializeDefautValue(f, e, defaultTableValue);
}
}
for (int iteration = 0; iteration < iterations; iteration++)
{
// count(e|f)
FETable count = new FETable();
// total(f)
Dictionary <string, float> total = new Dictionary <string, float>();
// s-total(e)
Dictionary <string, float> sTotal = new Dictionary <string, float>();
// Initialize default values
// - count
// - total
foreach (string f in trainingData.FWords)
{
total[f] = 0.0f;
foreach (string e in trainingData.EWords)
{
count.InitializeDefautValue(f, e, 0.0f);
}
}
foreach (SentencePair sentencePair in trainingData)
{
// Compute normalizations
foreach (string e in sentencePair.EWords)
{
sTotal[e] = 0.0f;
foreach (string f in sentencePair.FWords)
{
sTotal[e] += dictionary[f, e];
}
}
// Collect counts
foreach (string e in sentencePair.EWords)
{
foreach (string f in sentencePair.FWords)
{
count[f][e] += dictionary[f, e] / sTotal[e];
total[f] += dictionary[f, e] / sTotal[e];
}
}
}
// Esitmate probabilities
foreach (string f in trainingData.FWords)
{
foreach (string e in trainingData.EWords)
{
dictionary[f, e] = count[f][e] / total[f];
}
}
}
return(dictionary);
}