private OneDocumentInvindex StopWordsRemovals(OneDocumentInvindex doc)
{
OneDocumentInvindex DocumentUpdated;
List <string> documentTerms = new List <string>();
List <string> documentTermsTemp = doc.Terms;
for (int i = 0; i < documentTermsTemp.Count; i++)
{
if (!stop.Contains(documentTermsTemp[i]))
{
documentTerms.Add(documentTermsTemp[i]);
}
}
DocumentUpdated = new OneDocumentInvindex(doc.DocumentId, documentTerms, doc.Frequences, doc.Positions);
return(DocumentUpdated);
}
private void SaveInvertedIndex(OneDocumentInvindex doc)
{
for (int i = 0; i < doc.Terms.Count(); i++)
{
string term = doc.Terms.ElementAt(i);
conn.Open();
if (TermInserted(term, "GetTermDetails"))
{
UpdateTerm(term, doc.Positions[term], doc.Frequences[term].ToString(), doc.DocumentId.ToString());
}
else
{
InsertTerm(term, doc.Positions[term], doc.Frequences[term].ToString(), doc.DocumentId.ToString());
}
conn.Close();
}
}
private void HandlingModule1(object obj)
{
Dictionary <int, List <string> > index = (Dictionary <int, List <string> >)obj;
if (index != null)
{
semaphore.WaitOne();
saveWordToDataBase(index);
semaphore.Release();
List <string> stemmers = stemWord(index.Values.ElementAt(0));
Dictionary <string, int> frequences = Frequences(stemmers);
Dictionary <string, string> positions = Positions(stemmers);
OneDocumentInvindex doc = new OneDocumentInvindex(index.Keys.ElementAt(0), stemmers.Distinct().ToList(), frequences, positions);
semaphore.WaitOne();
SaveInvertedIndex(StopWordsRemovals(doc));
semaphore.Release();
}
}
