/// <summary>
/// Analyzes a message if it is or not SPAM.
/// Returns a boolean value for the validation.
/// </summary>
/// <param name="subject">The message subject.</param>
/// <param name="body">The mail message body.</param>
/// <param name="spamWordsFilename">The Spam Word List File.</param>
/// <param name="hamWordsFilename">The Ham Word List File.</param>
/// <param name="ignoreWordsFilename">The Ignore Word List File.</param>
/// <returns>True for SPAM, false if it isn't a SPAM.</returns>
static public bool AnalyzeMessage(string subject, string body, string spamWordsFilename, string hamWordsFilename, string ignoreWordsFilename)
{
// Load Spam Word List File
Hashtable SpamTab = new Hashtable();
Tokenizer.LoadFromFile(spamWordsFilename, ref SpamTab);
// Load Ham Word List File
Hashtable HamTab = new Hashtable();
Tokenizer.LoadFromFile(hamWordsFilename, ref HamTab);
// Load Ignore Word List File
Hashtable IgnoreTab = new Hashtable();
Tokenizer.LoadFromFile(ignoreWordsFilename, ref IgnoreTab);
//Parse Message Into Tokens
string[] msgTokens = Tokenizer.Parse(string.Format("{0} {1}", subject, body));
float I = 0;
float invI = 0;
foreach (string t in msgTokens)
{
if (!IgnoreTab.Contains(t))
{
float SpamCount = SpamTab.ContainsKey(t) ? (float)SpamTab[t] : 0f;
float HamCount = HamTab.ContainsKey(t) ? (float)HamTab[t] : 0f;
if (SpamCount == 0 && HamCount == 0)
{
continue;
}
// Calculate Probability
float bw = SpamCount / SpamTab.Count;
float gw = HamCount / HamTab.Count;
float pw = ((bw) / ((bw) + (gw)));
float s = 1f, x = .5f, n = SpamCount + HamCount;
float fw = ((s * x) + (n * pw)) / (s + n);
// Log Probability
I = I == 0 ? fw : I * fw;
invI = invI == 0 ? (1 - fw) : invI * (1 - fw);
}
}
//Calculate Prediction
float prediction = I / (I + invI);
if (prediction <= .45)
{
// No Spam
// Teach the Ham file based on the prediction
//Tokenizer.TeachListFile(hamWordsFilename, msgTokens, HamTab);
return(false);
}
else
if (prediction >= .55)
{
// Spam
// Teach the Spam file based on the prediction
//Tokenizer.TeachListFile(spamWordsFilename, msgTokens, SpamTab);
return(true);
}
// prediction > .45 && prediction < .55 - Unable to determine - by default no SPAM
// Teach the Ham file based on the prediction
//Tokenizer.TeachListFile(hamWordsFilename, msgTokens, HamTab);
return(false);
}
