private static void TestSimScore(string srcStr, string tarStr)
{
double score = GetScore(srcStr, tarStr);
int cost = EditDistance.GetEditDistance(srcStr, tarStr, 96, 90, 100, 94, 10, false);
Console.WriteLine(srcStr + "|" + tarStr + "|" + cost + "|" + score);
}
private static void TestEdSimScore(string srcStr, string tarStr)
{
double score = GetScore(srcStr, tarStr);
int cost = EditDistance.GetEditDistance(srcStr, tarStr, 96, 90, 100, 94, 10, false);
int penalty = OrthographicUtil.GetSplitPenalty(srcStr, tarStr, 90);
Console.WriteLine(srcStr + "|" + tarStr + "|" + cost + "|" + penalty + "|" + score);
}
public static double GetScore(string srcStr, string tarStr, int deleteCost, int insertCost, int replaceCost, int swapCost, int caseChangeCost, bool enhancedFlag, int splitCost)
{
int cost = EditDistance.GetEditDistance(srcStr, tarStr, deleteCost, insertCost, replaceCost, swapCost, caseChangeCost, enhancedFlag);
int penalty = OrthographicUtil.GetSplitPenalty(srcStr, tarStr, splitCost);
double score = OrthographicUtil.GetNormScore(cost + penalty, 1000.0);
return(score);
}
public static int GetDistance(string str1, string str2)
{
string str1Lc = str1.ToLower();
string str2Lc = str2.ToLower();
string str1Code = GetCode(str1Lc);
string str2Code = GetCode(str2Lc);
int dist = EditDistance.GetEditDistance(str1Code, str2Code);
return(dist);
}
public static string GetDistanceDetailStr(string str1, string str2)
{
string str1Lc = str1.ToLower();
string str2Lc = str2.ToLower();
string str1Code = GetCode(str1Lc);
string str2Code = GetCode(str2Lc);
int dist = EditDistance.GetEditDistance(str1Code, str2Code);
string detailStr = str1Code + GlobalVars.FS_STR + str2Code + GlobalVars.FS_STR + dist;
return(detailStr);
}
private static void TestPhoneticDistance(string srcStr, string tarStr)
{
int edDist = EditDistance.GetEditDistance(srcStr, tarStr, 95, 95, 100, 90, 10, false);
int pDist = GetDistance(srcStr, tarStr);
int maxCodeLength = 10;
string srcStrLc = srcStr.ToLower();
string tarStrLc = tarStr.ToLower();
//String srcM2 = Metaphone2.GetCode(srcStrLc, maxCodeLength);
//String tarM2 = Metaphone2.GetCode(tarStrLc, maxCodeLength);
string srcM2 = RefinedSoundex.GetCode(srcStrLc);
string tarM2 = RefinedSoundex.GetCode(tarStrLc);
Console.WriteLine(srcStr + "|" + tarStr + "|" + edDist + "|" + srcM2 + "|" + tarM2 + "|" + pDist);
}
// TBD: read the values from config file
public static double GetScore(string srcStr, string tarStr)
{
/// <summary>
/// init value form ensemble
/// int deleteCost = 95;
/// int insertCost = 95;
/// int replaceCost = 100;
/// int swapCost = 90;
///
/// </summary>
// new value
int deleteCost = 100;
int insertCost = 100;
int replaceCost = 100;
int swapCost = 100;
int caseChangeCost = 10;
bool enhancedFlag = false;
int splitCost = insertCost;
int maxCodeLength = 10;
/// <summary>
/// Test on different phonetic methods
/// String srcM2 = Metaphone2.GetCode(srcStr, maxCodeLength);
/// String tarM2 = Metaphone2.GetCode(tarStr, maxCodeLength);
/// String srcM2 = RefinedSoundex.GetCode(srcStr);
/// String tarM2 = RefinedSoundex.GetCode(tarStr);
/// String srcM2 = Caverphone2.GetCaverphone(srcStr);
/// String tarM2 = Caverphone2.GetCaverphone(tarStr);
/// String srcM2 = Metaphone.GetMetaphone(srcStr, maxCodeLength);
/// String tarM2 = Metaphone.GetMetaphone(tarStr, maxCodeLength);
/// Metaphone3 m3 = new Metaphone3();
/// m3.SetKeyLength(maxCodeLength);
/// String srcM2 = m3.GetMetaphone(srcStr);
/// String tarM2 = m3.GetMetaphone(tarStr);
///
/// </summary>
string srcM2 = Metaphone2.GetCode(srcStr, maxCodeLength);
string tarM2 = Metaphone2.GetCode(tarStr, maxCodeLength);
int cost = EditDistance.GetEditDistance(srcM2, tarM2, deleteCost, insertCost, replaceCost, swapCost, caseChangeCost, enhancedFlag);
int penalty = OrthographicUtil.GetSplitPenalty(srcStr, tarStr, splitCost);
double score = OrthographicUtil.GetNormScore(cost + penalty, 1000.0);
return(score);
}
public static int GetDistance(string srcStr, string tarStr)
{
int deleteCost = 1;
int insertCost = 1;
int replaceCost = 1; // penity on the replace
int swapCost = 1;
int caseChangeCost = 0;
bool enhancedFlag = false;
int splitCost = insertCost;
int maxCodeLength = 10;
string srcStrLc = srcStr.ToLower();
string tarStrLc = tarStr.ToLower();
//String srcM2 = Metaphone2.GetCode(srcStrLc, maxCodeLength);
//String tarM2 = Metaphone2.GetCode(tarStrLc, maxCodeLength);
string srcM2 = RefinedSoundex.GetCode(srcStrLc);
string tarM2 = RefinedSoundex.GetCode(tarStrLc);
int dist = EditDistance.GetEditDistance(srcM2, tarM2, deleteCost, insertCost, replaceCost, swapCost, caseChangeCost, enhancedFlag);
return(dist);
}