public static LogDebuggerControl getInstance()
{
if (_logDebugger == null)
_logDebugger = new LogDebuggerControl();
return _logDebugger;
}
public static LogDebuggerControl getInstance()
{
if (_logDebugger == null)
{
_logDebugger = new LogDebuggerControl();
}
return(_logDebugger);
}
/**
* This method returns the match level of the given wordlist according to a
* certain formula.
*/
public int getMatchLevel(String wordList, CategorizerOptions parameters)
{
char[] separators = { ' ', '\t', '\n' };
int numOfWords = Math.Max(1, wordList.Split(separators).Length);
int numOfKeywords = Math.Max(1, keywordList.Count);
int nonZero = 0;
double sumOfhistogram = 0;
double threshold = Math.Max(2.0, ((numOfWords * parameters.BETA) / numOfKeywords));
// keywordList and wordList are copied to a new arrays so that we won't change them(the originals)
List <String> keywordListCopied = new List <string>(keywordList);
String wordListCopied = (String)wordList.Clone();
// Transforming the keywordListCopied and wordListCopied to canonical form
//keywordListCopied.ForEach(canonicForm);
canonicForm(ref wordListCopied);
//wordListCopied = wordListCopied.ToLower();
int[] histogram = new int[numOfKeywords];
//Initialising the histogram array to zeros
for (int i = 0; i < numOfKeywords; i++)
{
histogram[i] = 0;
}
foreach (String keyword in keywordListCopied)
{
Regex objPattern = new Regex(keyword.ToLower());
int count = objPattern.Matches(wordListCopied, 0).Count;
int index = keywordListCopied.IndexOf(keyword);
if (count != 0 && histogram[index] == 0)
{
nonZero++;
}
if (histogram[index] < threshold)
{
int add = Math.Min(histogram[index] + count, (int)threshold) - histogram[index];
histogram[index] = histogram[index] + add;
sumOfhistogram = sumOfhistogram + add;
}
}
double nonZeroBonus = (nonZero * parameters.GAMMA) / numOfKeywords;
nonZeroBonus = Math.Min(parameters.NONZERO_MAX_EFFECT, nonZeroBonus);
double matchPercent = (sumOfhistogram * parameters.ALPHA) / numOfWords;
matchPercent = Math.Min(parameters.MATCH_MAX_EFFECT, matchPercent);
double total = parameters.MAX_MATCH_LEVEL * (nonZeroBonus + matchPercent) / (parameters.MATCH_MAX_EFFECT + parameters.NONZERO_MAX_EFFECT);
if (numOfWords < parameters.MIN_WORDS_LIMIT)
{
total = parameters.MIN_WORDS_PENLTY * total;
}
StreamWriter sw = null;
if (LogDebuggerControl.getInstance().debugCategorization)
{
if (!parameters.isRank)
{
sw = new
StreamWriter("_DEBUG_INFO_CATEGORIZER@" + System.Threading.Thread.CurrentThread.ManagedThreadId + ".txt", true);
sw.WriteLine(" ***** DATA FOR REQUEST ************************************************* ");
//sw.WriteLine(" .CONTENT WORDS: ");
//sw.WriteLine(wordListCopied.ToString());
sw.WriteLine(" .NUM OF WORDS: ");
sw.WriteLine(numOfWords.ToString());
sw.WriteLine(" .KEY WORDS: ");
sw.WriteLine(keywordListCopied.ToString());
sw.WriteLine(" .NUM OF KEY WORDS: ");
sw.WriteLine(numOfKeywords.ToString());
sw.WriteLine(" .THRESOLD PARAM: ");
sw.WriteLine(threshold.ToString());
sw.WriteLine(" .SUM OF HISTOGRAM: ");
sw.WriteLine(sumOfhistogram.ToString());
sw.WriteLine(" .NONZERO PARAM: ");
sw.WriteLine(nonZero.ToString());
sw.WriteLine(" .HISTOGRAM DATA:");
for (int j = 0; j < numOfKeywords; j++)
{
sw.WriteLine(" .[" + keywordList[j] + "] -> " + histogram[j].ToString());
}
sw.WriteLine(" .NON-ZERO BONUS: ");
sw.WriteLine(nonZeroBonus.ToString());
sw.WriteLine(" .MATCH PERCENT: ");
sw.WriteLine(matchPercent.ToString());
sw.WriteLine(" .TOTAL TRUST: ");
sw.WriteLine(total.ToString());
sw.WriteLine(" * END ****************************************************************** ");
sw.Close();
}
}
if (LogDebuggerControl.getInstance().debugCategorizationInRanker&& LogDebuggerControl.getInstance().debugRanker)
{
if (parameters.isRank)
{
sw = new StreamWriter("DataForRank" + System.Threading.Thread.CurrentThread.ManagedThreadId + ".txt", true);
sw.WriteLine(" ***** DATA FOR Categorizer ************************************************* ");
//sw.WriteLine(" .CONTENT WORDS: ");
//sw.WriteLine(wordListCopied.ToString());
sw.WriteLine(" .NUM OF WORDS: ");
sw.WriteLine(numOfWords.ToString());
//String[] wordListSplited = wordListCopied.Split(separators);
//for (int k = 0; k < numOfWords;k++ )
//{
// sw.WriteLine(" .[" + k + "] -> " + wordListSplited[k]);
//}
sw.WriteLine(" .KEY WORDS: ");
sw.WriteLine(keywordListCopied.ToString());
sw.WriteLine(" .NUM OF KEY WORDS: ");
sw.WriteLine(numOfKeywords.ToString());
sw.WriteLine(" .THRESOLD PARAM: ");
sw.WriteLine(threshold.ToString());
sw.WriteLine(" .SUM OF HISTOGRAM: ");
sw.WriteLine(sumOfhistogram.ToString());
sw.WriteLine(" .NONZERO PARAM: ");
sw.WriteLine(nonZero.ToString());
sw.WriteLine(" .HISTOGRAM DATA:");
for (int j = 0; j < numOfKeywords; j++)
{
sw.WriteLine(" .[" + keywordList[j] + "] -> " + histogram[j].ToString());
}
sw.WriteLine(" .NON-ZERO BONUS: ");
sw.WriteLine(nonZeroBonus.ToString());
sw.WriteLine(" .MATCH PERCENT: ");
sw.WriteLine(matchPercent.ToString());
sw.WriteLine(" .TOTAL TRUST: ");
sw.WriteLine(total.ToString());
sw.WriteLine(" * END ****************************************************************** ");
sw.Close();
}
}
return(Convert.ToInt32(total));
}
