public Dictionary <string, int> Learn(string languageCode, string sourcePath, string targetPath = null)
{
var text = Helper.GetFileContents(sourcePath);
var ngramBuilder = new NgramBuilder();
var ngrams = ngramBuilder.Get(text);
if (targetPath == null)
{
return(ngrams);
}
Save(languageCode, ngrams, targetPath);
return(ngrams);
}
public Dictionary <string, Dictionary <string, int> > Remember(string path)
{
var result = new Dictionary <string, Dictionary <string, int> >();
string line;
using (var reader = new StreamReader(path))
{
while ((line = reader.ReadLine()) != null)
{
var model = line.Split(':');
var ngramBuilder = new NgramBuilder();
result.Add(model[0], ngramBuilder.Load(model[1].Split('_')));
}
}
return(result);
}
public string Detect(string path, out int score)
{
score = 0;
var text = path; //read file
var ngramBuilder = new NgramBuilder(MaxLength, true);
var ngrams = ngramBuilder.Get(text); //create an ngram dictionary
if (ngrams == null)
{
return(null);
}
var shortestDistance = int.MaxValue;
var probability = 0;
string lowestScoringLanguage = null;
foreach (var availableLanguage in _availableLanguages)
{
//calculate distance between language and ngrams
var distance = 0;
var probabilityHits = 0;
foreach (var ngram in ngrams)
{
if (availableLanguage.Value.ContainsKey(ngram.Key))
{
distance += ngram.Value - availableLanguage.Value[ngram.Key];
probabilityHits++;
}
else
{
distance += MaxPenalty;
}
if (distance > shortestDistance)
{
break;
}
}
if (distance < shortestDistance)
{
shortestDistance = distance;
lowestScoringLanguage = availableLanguage.Key;
probability = probabilityHits;
}
}
probability = probability * 4;
if (probability > 100)
{
probability = 100;
}
score = probability;
return(lowestScoringLanguage);
}