/// <summary>
/// THIS SMAP LOOP FOR TFIDF Algorithm
/// </summary>
/// <param name="files"> all the email files</param>
/// <param name="limit"> how many emails </param>
/// <param name="SPAM"> are they spam or not</param>
/// <param name="obj"> the TFIDF object </param>
private void Loop(string[] files, decimal limit, bool SPAM, TFIDF obj)
{
try
{
for (int i = 0; i < (int)limit; i++)
{
StreamReader myFile = new StreamReader(files[i]);
string allFile = myFile.ReadToEnd();
string[] words = allFile.Split(' ');
obj.addDocument(words, SPAM); // add the document and its words
}
}
catch
{
}
}
/// <summary>
/// this calcuate the shown statics for each algorithm
/// </summary>
/// <param name="hams">list of ham files </param>
/// <param name="spams">list of spam files </param>
/// <param name="algorithm"></param>
/// <returns></returns>
staticSet calculateStatics(string[] hams, string[] spams, AlgorithmBox algorithm)
{
staticSet myStatics = new staticSet(0);
int hamScores = 0;
int spamScores = 0;
myStatics.totalham = (float)hamNumber.Value;
myStatics.totalspam = (float)spamNumber.Value;
progressBar.Value = 0;
progressBar.Maximum = (int)(myStatics.totalham + myStatics.totalspam);
DateTime tic = DateTime.Now;
Algorithm myAlgo = null;
switch (algorithm)
{
case AlgorithmBox.LCS:
myAlgo = new LCSWord();
break;
case AlgorithmBox.Levenshtein:
myAlgo = new Levenshtien();
break;
case AlgorithmBox.Jaro:
myAlgo = new Jaro();
break;
case AlgorithmBox.JaroWinker:
myAlgo = new JaroWinker();
break;
case AlgorithmBox.BiGram:
myAlgo = new BGram();
break;
default:
break;
}
// we need a conditon here
if (selectedAlgorithm == AlgorithmBox.TFIDF)
{
TFIDF myTFI = new TFIDF();
myTFI.Terms = phrasePower;
Loop(hams, hamNumber.Value, false, myTFI);
Loop(spams, spamNumber.Value, true, myTFI);
// the TFIDF runner
myTFI.run();
float threshold = ((float)thresholdValue.Value) / 100F;
myTFI.claculate(ref hamScores, ref spamScores, threshold);
}
else
{
// other runners
hamScores = Loop(hams, hamNumber.Value, myAlgo);
spamScores = Loop(spams, spamNumber.Value, myAlgo);
}
DateTime toc = DateTime.Now;
TimeSpan timeSpent = toc - tic;
// calculate the spent time , a simple timing mechanism
myStatics.time = (float)timeSpent.TotalSeconds;
if (myStatics.totalspam > 0)
{
// false negatives
myStatics.fnegatives = (myStatics.totalspam - spamScores) / myStatics.totalspam;
// spam recall
myStatics.spamrecall = spamScores / myStatics.totalspam;
// spam precision
myStatics.spamper = spamScores / (spamScores + (myStatics.totalham - hamScores));
}
if (myStatics.totalham > 0)
{
// false positives
myStatics.fpositives = (myStatics.totalham - hamScores) / myStatics.totalham;
}
if (myStatics.totalham > 0 || myStatics.totalspam > 0)
{
// accuracy
myStatics.accuracy = 1 - (((myStatics.totalspam - spamScores) + (myStatics.totalham - hamScores)) / (myStatics.totalham + myStatics.totalspam));
}
// Spam recall
return(myStatics);
}
