//Calculate cosin and BM fields
public void calculateWeight()
{
for (int i = 0; i < r.files.Count; i++)
{
Debug.WriteLine(i);
//Run on all files again
Dictionary <string, string> d = r.ProccessDocs(r.files[i]);
foreach (string docNo in d.Keys)
{
//Parse DOC
Dictionary <string, termInfo> docdic = p.Parse(d[docNo]);
//length dictionary
docslength[docNo.Trim(' ')] = docdic.Count;
//Wight dictionary
DocWeight.Add(docNo.Trim(' '), 0);
//max tf dic
Docsmaxtf.Add(docNo.Trim(' '), p.maxtf);
foreach (string trm in docdic.Keys)
{
if (docdic.ContainsKey(trm) && ind.dic.ContainsKey(trm))
{ //For Cosin
double W = (((double)docdic[trm].tf / (double)p.maxtf) * (double)ind.dic[trm].idf);
DocWeight[docNo.Trim(' ')] += (W * W);
}
}
DocWeight[docNo.Trim(' ')] = Math.Sqrt(DocWeight[docNo.Trim(' ')]);
}
}
//for BM Formula
foreach (string s in docslength.Keys)
{
avgDL += docslength[s];
}
avgDL = avgDL / (double)docslength.Count;
}
private void StartEngine()
{
//Delete Exists Directory
if (Directory.Exists(pathclose + @"\AfterPost"))
{
Directory.Delete(pathclose + @"\AfterPost", true);
}
if (Directory.Exists(pathclose + @"\tempPost"))
{
Directory.Delete(pathclose + @"\tempPost", true);
}
/* if (Directory.Exists(pathclose + @"\CacheDic"))
* Directory.Delete(pathclose + @"\CacheDic", true);*/
//initlizing reader,parser,indexer
r = new ReadFile(pathopen);
p = new Parser(pathopen + @"\stop_words.txt", isStem);
ind = new Indexer(pathclose, isStem);
//For program timing
DateTime start = DateTime.Now;
for (int i = 0; i < r.files.Count; i++)
{
Debug.WriteLine(i);
//return dictionary<DocNO, TEXT>
d = r.ProccessDocs(r.files[i]);
foreach (string s in d.Keys)
{
//Parse DOC
Dictionary <string, termInfo> docdic = p.Parse(d[s]);
//make temp Post file
ind.tempPosting(docdic, s.Trim(' '), p.maxterm, p.maxtf);
}
}
StreamWriter sw = new StreamWriter(pathopen + @"\docsMap.txt");
sw.WriteLine(r.sb.ToString());
sw.Close();
ind.tempPost(); //Write the last dictionary
p.clearSteam(); // cleans the stemmers dictionary
ind.mergefile(); // merge and split
ind.writeTextDic(); // WriteDic For show
TimeSpan ts3 = DateTime.Now - start;
//The requested popup of this run
System.Windows.Forms.MessageBox.Show("Number Of Docs Indexed : " + ind.TotalDoc + "\nTime Of Running : " + ts3.TotalSeconds + "\nIndex Size[bytes] : " + ind.PostSize + "\nCache Size[bytes] : " + ind.cacheSize, "Done!", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
