public void HitungPembobotanKata()
{
List <Term> InvertedIndex = Lingkungan.LoadInvertedIndex();
if (InvertedIndex == null)
{
InvertedIndex = new List <Term>();
}
if (db == null)
{
db = new dbDataContext();
}
List <tbDomain> Domain = db.tbDomains.ToList(); // seluruh domain yang ada
List <int> DomainCount = new List <int>(); // Counter setiap domain
for (int i = 0; i < Domain.Count; i++) // inisialisasi counter domain
{
DomainCount.Add(0);
}
// Hitung jumlah kata dalam seluruh dokumen dan perdomain
int TermCount = 0; //counter kata
foreach (var item in InvertedIndex)
{
TermCount += item.Index.Count;
for (int i = 0; i < Domain.Count; i++)
{
DomainCount[i] += item.Index.Where(x => x.DomainID == Domain[i].Id).Count();
}
}
for (int i = 1; i < DomainCount.Count; i++)
{
if (DomainCount[i] == 0)
{
for (int j = 1; j < DomainCount.Count; j++)
{
DomainCount[i]++;
}
break;
}
}
//pembobotan
foreach (var item in InvertedIndex)
{
item.Bobot.Clear();
for (int i = 0; i < Domain.Count; i++)
{
int count = item.Index.Where(x => x.DomainID == Domain[i].Id).Count();
double bobot = Math.Log10((Lingkungan.getLambda(0) * (count / TermCount)) +
(Lingkungan.getLambda(1) * count / DomainCount[i]));
item.Bobot.Add(bobot);
}
}
Lingkungan.SaveInvertedIndex(InvertedIndex);
}
public tbInfDetail[] PencarianInformasi(Dialogue inpt, int domain, List <Term> extra)
{
List <tbInfDetail> dataDomain = db.tbInfDetails.Where(x => x.tbInformasi.DomainID == domain).ToList();
List <decimal> bobot = new List <decimal>();
List <Term> qry = inpt.StringToTerm();
if (extra != null)
{
foreach (var item in extra)
{
if (qry.Where(x => x.Word == item.Word).FirstOrDefault() == null)
{
qry.Add(item);
}
}
}
int TermCount = 0;
foreach (var item in Lingkungan.LoadInvertedIndex())
{
TermCount += item.Index.Count;
}
decimal termC = (decimal)TermCount;
decimal b;
for (int i = 0; i < dataDomain.Count; i++)
{
string[] fragment = dataDomain[i].info.Split(' ').Where(x => string.IsNullOrWhiteSpace(x) == false).ToArray();
b = 1;
foreach (var item in qry)
{
decimal FragmentTermcounter = Convert.ToDecimal(fragment.Where(x => x.ToLower().Contains(item.Word.ToLower())).Count());
decimal FragmentCounter = Convert.ToDecimal(fragment.Count());
decimal itemCounter = Convert.ToDecimal(item.Bobot.Count);
b = b * ((FragmentTermcounter / FragmentCounter) * Convert.ToDecimal(Lingkungan.getLambda(1)) +
(itemCounter / termC) * Convert.ToDecimal(Lingkungan.getLambda(0)));
}
bobot.Add(b);
}
List <tbInfDetail> top10 = new List <tbInfDetail>();
while (dataDomain.Count > 0 && top10.Count < 10)
{
int idx = bobot.LastIndexOf(bobot.Max());
top10.Add(dataDomain[idx]);
bobot.RemoveAt(idx);
dataDomain.RemoveAt(idx);
}
return(top10.ToArray());
}