public void ExtractCognatesAndFalseFriends(string outputFileName)
{
StreamWriter writer = new StreamWriter(outputFileName, false, FILE_ENCODING);
using (writer)
{
writer.WriteLine("BG Word;RU Word;Friends?;MMED Similarity;BG Count;RU Count;Corresponding Count;Friendness;Semantic Similarity");
foreach (string bgWord in this.allBgWords.Keys)
{
Console.WriteLine("Extracting words statistics: " + bgWord);
foreach (string ruWord in this.allRuWords.Keys)
{
double wordsSimilarity = MMEDR.CalculateSimilarity(bgWord, ruWord);
if (wordsSimilarity >= MIN_WORD_SIMILARITY)
{
string friendsJudgeAnswer = GetFriendsJudgeAnswer(bgWord, ruWord);
if (friendsJudgeAnswer != null)
{
float bgCount = CalculateBgCount(bgWord);
float ruCount = CalculateRuCount(ruWord);
float correspondingCount = CalculateCorrespondingCount(bgWord, ruWord);
float friendness = CalculateFriendness(bgCount, ruCount, correspondingCount);
double semanticSimilarity = SemanticSimilarityUtils.CrossSim(bgWord, ruWord);
writer.WriteLine("{0};{1};{2};{3};{4};{5};{6};{7};{8}",
bgWord, ruWord, friendsJudgeAnswer, wordsSimilarity, bgCount, ruCount,
correspondingCount, friendness, semanticSimilarity);
writer.Flush();
}
}
}
}
}
}
private static double CalcMMEDR(string bgWord, string ruWord)
{
// Transcribe the Bulgarian word
ReplaceAll(ref bgWord, "щ", "шт");
ReplaceAll(ref bgWord, "ьо", "ё");
ReplaceAll(ref bgWord, "йо", "ё");
// Transcribe the Russian word
ReplaceAll(ref ruWord, "э", "е");
ReplaceAll(ref ruWord, "щ", "сч");
ReplaceAll(ref ruWord, "ъ", "");
ReplaceAll(ref ruWord, "ы", "и");
ReplaceAll(ref ruWord, "ь", "");
// Remove double consonants in the Bulgarian word
ReplaceAll(ref bgWord, "зс", "с");
// Remove double consonants in the Russian word
ReplaceAll(ref ruWord, "бб", "б");
ReplaceAll(ref ruWord, "жж", "ж");
ReplaceAll(ref ruWord, "кк", "к");
ReplaceAll(ref ruWord, "лл", "л");
ReplaceAll(ref ruWord, "мм", "м");
ReplaceAll(ref ruWord, "пп", "п");
ReplaceAll(ref ruWord, "сс", "с");
ReplaceAll(ref ruWord, "фф", "ф");
// Calculate weighted Levenstain distance (MMEDR)
MMEDR distCalc = new MMEDR(bgWord, ruWord);
double distance = distCalc.CalcDistance(bgWord.Length, ruWord.Length);
double maxLen = Math.Max(bgWord.Length, ruWord.Length);
double similarity = (maxLen - distance) / maxLen;
return(similarity);
}
