public static string Classify(string snippet, out Dictionary <string, double> scores)
{
if (_instance == null)
{
_instance = new CodeClassifier();
}
scores = new Dictionary <string, double>();
List <Token> tokens = GetAllTokens(snippet);
double maxScore = 0;
string bestMatchLanguage = null;
foreach (MatchTree matchTree in _matchTrees)
{
double score = 0;
for (int index = 0; index < tokens.Count; index++)
{
score += ScoreTokens(matchTree.MatchTreeRoot, tokens, index);
}
score = score / tokens.Count() / matchTree.TotalPossibleScore;
//Console.WriteLine(matchTree.Language + "\t" + score);
scores.Add(matchTree.Language, score);
if (score > maxScore)
{
maxScore = score;
bestMatchLanguage = matchTree.Language;
}
}
return(bestMatchLanguage);
}
public static string Classify(string snippet, out Dictionary<string, double> scores )
{
if (_instance == null)
{
_instance = new CodeClassifier();
}
scores = new Dictionary<string, double>();
List<Token> tokens = GetAllTokens(snippet);
double maxScore = 0;
string bestMatchLanguage = null;
foreach (MatchTree matchTree in _matchTrees)
{
double score = 0;
for (int index = 0; index < tokens.Count; index++)
{
score += ScoreTokens(matchTree.MatchTreeRoot, tokens, index);
}
score = score / tokens.Count() / matchTree.TotalPossibleScore;
//Console.WriteLine(matchTree.Language + "\t" + score);
scores.Add(matchTree.Language, score);
if (score > maxScore)
{
maxScore = score;
bestMatchLanguage = matchTree.Language;
}
}
return bestMatchLanguage;
}
public static string Classify(string snippet, out double certainty, out Dictionary <string, double> scores)
{
if (_instance == null)
{
_instance = new CodeClassifier();
}
Dictionary <string, double> scoresTp;
string bestLanguageTp = ClassifyByTokenProbability(snippet, out scoresTp);
Console.WriteLine("\n\nToken frequencies: ");
OutputLanguageScores(scoresTp, bestLanguageTp);
double certTp = CalculateCertainty(scoresTp);
Dictionary <string, double> scoresMt;
string bestLanguageMt = ClassifyByMatchTrees(snippet, out scoresMt);
Console.WriteLine("\n\nMatch trees: ");
OutputLanguageScores(scoresMt, bestLanguageMt);
double certMt = CalculateCertainty(scoresMt);
scores = new Dictionary <string, double>();
foreach (string language in scoresTp.Keys.ToList())
{
scores[language] = scoresTp[language] * certTp + scoresMt[language] * certMt;
}
certainty = CalculateCertainty(scores);
return(CalculateWinningLanguage(scores));
//if (certTp > certMt)
//{
// scores = scoresTp;
// certainty = certTp;
// return bestLanguageTp;
//}
//else
//{
// scores = scoresMt;
// certainty = certMt;
// return bestLanguageMt;
//}
}
public static string Classify(string snippet, out double certainty, out Dictionary<string, double> scores)
{
if (_instance == null)
{
_instance = new CodeClassifier();
}
Dictionary<string, double> scoresTp;
string bestLanguageTp = ClassifyByTokenProbability(snippet, out scoresTp);
Console.WriteLine("\n\nToken frequencies: ");
OutputLanguageScores(scoresTp, bestLanguageTp);
double certTp = CalculateCertainty(scoresTp);
Dictionary<string, double> scoresMt;
string bestLanguageMt = ClassifyByMatchTrees(snippet, out scoresMt);
Console.WriteLine("\n\nMatch trees: ");
OutputLanguageScores(scoresMt, bestLanguageMt);
double certMt = CalculateCertainty(scoresMt);
scores = new Dictionary<string, double>();
foreach (string language in scoresTp.Keys.ToList())
{
scores[language] = scoresTp[language]*certTp + scoresMt[language]*certMt;
}
certainty = CalculateCertainty(scores);
return CalculateWinningLanguage(scores);
//if (certTp > certMt)
//{
// scores = scoresTp;
// certainty = certTp;
// return bestLanguageTp;
//}
//else
//{
// scores = scoresMt;
// certainty = certMt;
// return bestLanguageMt;
//}
}
