/// <summary>
/// Calculates the distance between the two language distribution,
/// given that <code>text</code> is the LetterDistribution based on text.
/// </summary>
/// <param name="language"></param>
/// <param name="text"></param>
/// <returns></returns>
public static double Distance(LetterDistribution language, LetterDistribution text)
{
double distance2 = 0;
foreach (char key in text.letters.Keys)
{
double textFrequency = (double)(text.letters[key]);
double languageFrequency = (double)(language.letters[key]);
double sub = textFrequency - languageFrequency;
distance2 += sub * sub;
}
return(distance2);
}
/// <summary>
/// Detects the source language based on the letter frequency.
/// </summary>
/// <param name="text"></param>
/// <returns>The language code (e.g. "EN" for English), or null if no language was detected</returns>
public static string detect(string text)
{
LetterDistribution textDistribution = new LetterDistribution(text);
double minDistance = double.MaxValue;
KeyValuePair <string, LetterDistribution> pair = new KeyValuePair <string, LetterDistribution>(null, null);
foreach (var lang in languages)
{
double distance = LetterDistribution.Distance(lang.Value, textDistribution);
if (distance < minDistance)
{
pair = lang;
minDistance = distance;
}
}
return(pair.Key);
}
