private void button3_Click(object sender, EventArgs e)
{
//bbl = new BoardBoolean(ListDocuments);
//bbl.CreateBoardVobu();
//bbl.CreateBoardDictionary();
var dt = new DataTable();
dt.Columns.Add("Từ Khóa");
dt.Columns.Add("Chỉ số tài liệu");
dt.Columns.Add("Tần số");
dt.Columns.Add("TF");
dt.Columns.Add("IDF");
dt.Columns.Add("TF * IDF");
foreach (var tuple in bbl.GetDictionList())
{
dt.Rows.Add(tuple.Key, tuple.Value.Item1, tuple.Value.Item2, "", tuple.Value.Item3.ToString(CultureInfo.InvariantCulture));
foreach (var variable in tuple.Value.Item4)
{
dt.Rows.Add("", variable.Item1, variable.Item2, variable.Item3.ToString(CultureInfo.InvariantCulture), "", variable.Item4.ToString(CultureInfo.InvariantCulture));
}
}
var dicAndPosting = new DicAndPosting(dt);
dicAndPosting.Show();
}
private void button2_Click(object sender, EventArgs e)
{
var bbl = new BoardBoolean(ListDocuments);
bbl.CreateBoardVobu();
var lstBoard = bbl.Vobu;
foreach (var tuple in lstBoard)
{
Console.WriteLine(tuple.Item1 + "\t" + tuple.Item2 + "\t" + tuple.Item3);
}
Console.WriteLine("-------------------------------------------------------");
bbl.CreateBoardDictionary();
foreach (var tuple in bbl.GetDictionList())
{
Console.WriteLine(tuple.Key + "\t" + tuple.Value.Item1 + "\t" + tuple.Value.Item2);
foreach (var variable in tuple.Value.Item4)
{
Console.WriteLine("\t" + variable.Item1 + "\t" + variable.Item2);
}
}
}
public List <KeyValuePair <string, double> > FileResult()
{
var dic = _boolean.GetDictionList();
var lst = Query.Lstq;
//tên từ, Tuple<chỉ số tài liêu, tần số, idf.<List<tuple<tài liệu,tần số,tf,tf*idf>>>
IDictionary <string, Tuple <int, int, double, List <Tuple <string, int, float, double> > > > listItem = new Dictionary <string, Tuple <int, int, double, List <Tuple <string, int, float, double> > > >();
Tuple <int, int, double, List <Tuple <string, int, float, double> > > Item;
var lstQuery = new List <Tuple <string, double> >();
//query: tu, tf
lst = lst.Distinct().ToList();
foreach (var item in lst)
{
if (!dic.ContainsKey(item.Item1))
{
continue;
}
Item = dic.First(a => a.Key == item.Item1).Value;
//tu, tf*idf
lstQuery.Add(new Tuple <string, double>(item.Item1, item.Item2 * dic[item.Item1].Item3));// tu ,tf
listItem.Add(item.Item1, Item);
}
//tai lieu, tu, tf*idf
var lstFile = (from item in listItem from subitem in item.Value.Item4
select new Tuple <string, string, double>(subitem.Item1, item.Key, subitem.Item4))
.OrderBy(a => a.Item1).ToList();
// gop lai
var dicFile = new Dictionary <string, List <Tuple <string, double> > >();
foreach (var item in lstFile)
{
if (dicFile.ContainsKey(item.Item1) == true)
{
dicFile[item.Item1].Add(new Tuple <string, double>(item.Item2, item.Item3));
}
else
{
var tup = new Tuple <string, double>(item.Item2, item.Item3);
dicFile.Add(item.Item1, new List <Tuple <string, double> >(new Tuple <string, double>[] { tup }));
}
}
var docTfidf = new Dictionary <string, double>();
foreach (var item in dicFile)
{
var doDaiTaiLieu = Math.Sqrt(item.Value.Sum(a => Math.Pow(a.Item2, 2)));
var file = item.Value.ToList();
var d = 0.0;
var x = item.Value.Sum(subitem => lstQuery.FirstOrDefault(a => a.Item1 == subitem.Item1).Item2 *subitem.Item2);
var doDaiQuery = Math.Sqrt(lstQuery.Sum(e => Math.Pow(e.Item2, 2)));
var c = (float)((float)x / ((float)doDaiTaiLieu * (float)doDaiQuery));
docTfidf.Add(item.Key, (double)c);
}
var listFileSortByFre = docTfidf.ToList();
return(listFileSortByFre);
//tai lieu,tu,tf*idf
//new Dictionary<string,List<Tuple<string,double>>>();
//var listFileSortByFre = dicFile.ToList().OrderByDescending(a => a.Value.Item2).ThenBy(a => a.Value.Item1).Where(a => a.Value.Item1 >= 3);
//return listFileSortByFre;
}