Beispiel #1
0
 public TermParser()
 {
     stopList    = new d.StopList();
     delims      = parseHelper.GetDelims();
     stemmer     = new PorterStemmer();
     termData    = new d.TermData();
     DocData     = new d.DocData();
     termdocData = new d.TermDocData();
     parseHelper = new ParseHelper();
 }
Beispiel #2
0
        //watch for divide by 0
        public void CalculateIdfsA()
        {
            double docCount = Convert.ToDouble(new d.DocData().GetCount()); //required for correct division

            d.TermData td = new d.TermData();
            DataTable  dt = td.GetDocCountsA();

            foreach (DataRow dr in dt.Rows)
            {
                td.UpdateIdfA(dr[0].ToString(), Convert.ToSingle(Math.Log(docCount / (int)dr[1], 2)));
            }
        }
Beispiel #3
0
        public void ConvertTermTable()
        {
            DataTable dt = new d.TermData().GetAll();
            string    term;
            string    docccount;
            string    doccount_a;
            string    idf;
            string    idf_a;
            string    totalcount;
            string    totalcount_w;
            string    totalcount_a;
            string    totalcount_wa;

            foreach (DataRow dr in dt.Rows)
            {
                StringBuilder sb = new StringBuilder();
                term          = dr[0].ToString();
                docccount     = dr[1].ToString();
                doccount_a    = dr[2].ToString();
                idf           = dr[3].ToString();
                idf_a         = dr[4].ToString();
                totalcount    = dr[5].ToString();
                totalcount_w  = dr[6].ToString();
                totalcount_a  = dr[7].ToString();
                totalcount_wa = dr[8].ToString();
                sb.AppendFormat("{0} ", term);
                sb.AppendFormat("{0} ", docccount);
                sb.AppendFormat("{0} ", doccount_a);
                sb.AppendFormat("{0} ", idf);
                sb.AppendFormat("{0} ", idf_a);
                sb.AppendFormat("{0} ", totalcount);
                sb.AppendFormat("{0} ", totalcount_w);
                sb.AppendFormat("{0} ", totalcount_a);
                sb.AppendFormat("{0} ", totalcount_wa);
                Console.WriteLine(sb.ToString());
            }
        }