public void getDoc_MengandungQuery(string[] query)
{
double[] hasilPembilang_Perkata;
double[] hasilPenyebut_Perkata;
for (int i = 0; i < query.Length; i++)
{
string queryPakek = cariID_Term(query[i]);
if (queryPakek.Equals("0"))
{
Console.WriteLine("Tidak ada dokument yang memakai kata '" + query + "'");
Console.WriteLine("======================================================================");
}
else
{
koneksi con = new koneksi();
con.openConnection();
string queryDF = "SELECT \"idDokumen\", \"id_Term\", \"TF-IDF\" FROM public.\"bobotTerm\" where" +
" \"id_Term\"='" + queryPakek + "'";
DataTable result = con.getResult(queryDF);
string[] dataID_Term = new string[result.Rows.Count];
hasilTF_IDF_Perkata_Pembagi = new double[dataID_Term.Length];
hasilTF_IDF_Perkata_Pembilang = new double[dataID_Term.Length];
string[] dataTF_IDF_Term = new string[result.Rows.Count];
}
}
}
public void simpanTermBaru(string term)
{
con.openConnection();
this.con = new koneksi();
string queryTerm = "INSERT INTO public.\"Term\"(\"Term\") VALUES ('" + term + "');";
con.excequteQuery(queryTerm);
con.stopAccess();
}
public void updateDF(string term, string ndata)
{
con.openConnection();
this.con = new koneksi();
string queryDF = "UPDATE public.\"Term\" set \"DF\"=" + ndata + " WHERE \"Term\"='" + term + "';";
con.excequteQuery(queryDF);
con.stopAccess();
}
public void insert_ToBobot(string idDoc, string idTerm, int tf)
{
this.con = new koneksi();
con.openConnection();
string queryInsert_Dokument = "INSERT INTO public.\"bobotTerm\"(\"idDokumen\", \"id_Term\", \"Tf\") VALUES ('" + idDoc + "', '" + idTerm + "', " + tf + ");";
con.excequteQuery(queryInsert_Dokument);
con.stopAccess();
}
public int getCOunt()
{
this.con = new koneksi();
string query = "SELECT COUNT(*) FROM public.katadasar";
DataTable result = this.con.getResult(query);
int count = result.Rows.Count;
return(count);
}
public void updateTF_IDF(double value, string idBobotTerm)
{
con.openConnection();
this.con = new koneksi();
string queryTF_IDF = "UPDATE public.\"bobotTerm\" SET \"TF-IDF\"='" + value + "' WHERE \"id_bobotTerm\"='" + idBobotTerm + "';";
con.excequteQuery(queryTF_IDF);
con.stopAccess();
}
public string getValueDF(string term)
{
string valueCari = null;
this.con = new koneksi();
con.openConnection();
string termMasuk = "SELECT \"DF\" FROM public.\"Term\" where \"Term\"='" + term + "';";
DataTable result = this.con.getResult(termMasuk);
valueCari = result.Rows[0]["DF"].ToString();
con.stopAccess();
return(valueCari);
}
public string cariID_Term(string data)
{
string termCari = null;
this.con = new koneksi();
con.openConnection();
string termMasuk = "SELECT * FROM public.\"Term\" where \"Term\"='" + data + "';";
DataTable result2 = this.con.getResult(termMasuk);
termCari = result2.Rows[0]["id_Term"].ToString();
con.stopAccess();
return(termCari);
}
public string getIDF_Term(string term)
{
koneksi con = new koneksi();
con.openConnection();
double getIDF_Term = 0;
string QgetDF_Term = "SELECT \"IDF\" FROM public.\"Term\" where \"Term\"='" + term + "';";
DataTable result = con.getResult(QgetDF_Term);
getIDF_Term = Convert.ToDouble(result.Rows[0]["IDF"]);
con.stopAccess();
return(getIDF_Term.ToString());
}
public string[] getId_Doc(string idTerm)
{
koneksi con = new koneksi();
con.openConnection();
string query = "SELECT \"idDokumen\" FROM public.\"bobotTerm\" where \"id_Term\" = '" + idTerm + "'";
DataTable result = con.getResult(query);
string[] arrayDoc = new string[result.Rows.Count];
for (int i = 0; i < result.Rows.Count; i++)
{
arrayDoc[i] = result.Rows[i]["idDokumen"].ToString();
}
return(arrayDoc);
}
public string[] getID_Term(string[] query)
{
koneksi con = new koneksi();
con.openConnection();
string queryTerm = "SELECT \"id_Term\" FROM public.\"Term\" where \"Term\" in ( " + string.Join(", ", query) + ");";
DataTable result = con.getResult(queryTerm);
string[] arrayTerm = new string[result.Rows.Count];
for (int i = 0; i < result.Rows.Count; i++)
{
arrayTerm[i] = "'" + result.Rows[i]["id_Term"] + "'".ToString();
}
return(arrayTerm);
}
public void setIDF()
{
con.openConnection();
this.con = new koneksi();
string queryDF = "SELECT \"id_Term\" FROM public.\"Term\";";
DataTable result = this.con.getResult(queryDF);
string[] dataID_Term = new string[result.Rows.Count];
for (int i = 0; i < result.Rows.Count; i++)
{
dataID_Term[i] = result.Rows[i]["id_Term"].ToString();
double value = count_IDF(dataID_Term[i]);
updateIDF(value, dataID_Term[i]);
}
con.stopAccess();
}
public bool getKataSambung(string kata)
{
this.con = new koneksi();
string query = "SELECT kata FROM public.kata_sambung where kata = '" + kata + "'";
DataTable result = this.con.getResult(query);
con.closeConnection();
if (result.Rows.Count == 1)
{
return(true);
}
else
{
return(false);
}
}
public void get_LongDoc()
{
con.openConnection();
this.con = new koneksi();
string queryDF = "select count(distinct \"idDokumen\") as getLong from public.\"bobotTerm\";";
DataTable result = this.con.getResult(queryDF);
getNumber = Convert.ToInt32(result.Rows[0]["getLong"].ToString());
if (getNumber == 0)
{
getNumber = 1;
}
else
{
getNumber += 1;
}
con.stopAccess();
}
public bool cekID_Doc(string id)
{
bool status = false;
this.con = new koneksi();
con.openConnection();
string cek_KesediaanID = "SELECT * FROM public.\"bobotTerm\" where \"idDokumen\"='" + id + "';";
DataTable result = this.con.getResult(cek_KesediaanID);
if (result.Rows.Count > 0)
{
status = false;
}
else
{
status = true;
}
return(status);
}
public string cariID_Term(string data)
{
koneksi con = new koneksi();
con.openConnection();
string termCari = null;
string termMasuk = "SELECT * FROM public.\"Term\" where \"Term\"='" + data + "';";
DataTable result2 = con.getResult(termMasuk);
idDoc = new string[result2.Rows.Count];
if (result2.Rows.Count >= 1)
{
termCari = result2.Rows[0]["id_Term"].ToString();
con.stopAccess();
return(termCari);
}
else
{
return("0");
}
}
public double cariTF_IDF(string id_doc, string idTerm)
{
koneksi con = new koneksi();
con.openConnection();
double termCari;
string termMasuk = "SELECT \"TF-IDF\" FROM public.\"bobotTerm\" where" +
" \"idDokumen\"=" + id_doc + " and \"id_Term\"=" + idTerm + ";";
DataTable result2 = con.getResult(termMasuk);
if (result2.Rows.Count >= 1)
{
termCari = Convert.ToDouble(result2.Rows[0]["TF-IDF"]);
con.stopAccess();
return(termCari);
}
else
{
return(0);
}
}
public bool runVSM_DB(string data, string id, int tf)
{
string idTerm;
this.con = new koneksi();
con.openConnection();
string cek_KesediaanTerm = "SELECT * FROM public.\"Term\" where \"Term\"='" + data + "';";
DataTable result = this.con.getResult(cek_KesediaanTerm);
con.stopAccess();
if (result.Rows.Count == 0)
{
//simpan Term baru
con.openConnection();
this.con = new koneksi();
string queryTerm = "INSERT INTO public.\"Term\"(\"Term\", \"DF\") VALUES ('" + data + "','1');";
con.excequteQuery(queryTerm);
con.stopAccess();
con.closeConnection();
//baca idTermnya
idTerm = cariID_Term(data);
insert_ToBobot(id, idTerm, tf);
con.stopAccess();
con.closeConnection();
}
else // ketika sama
{
//baca idTermnya
idTerm = cariID_Term(data);
//update nilai df kata-n
int dfBefore = Convert.ToInt32(getValueDF(data));
int data_update = (dfBefore + 1);
updateDF(data, Convert.ToString(data_update));
insert_ToBobot(id, idTerm, tf);
con.stopAccess();
}
//simpan dokument
return(true);
}
public string[] getID_Doc(string[] idTerm)
{
koneksi con = new koneksi();
con.openConnection();
string queryTerm = "SELECT \"idDokumen\" FROM public.\"bobotTerm\" where \"id_Term\" in ( " + string.Join(", ", idTerm) + " );";
DataTable result = con.getResult(queryTerm);
string[] arrayDoc = new string[result.Rows.Count];
List <string> List_idDoc = new List <string>();
for (int i = 0; i < result.Rows.Count; i++)
{
if (!List_idDoc.Contains(result.Rows[i]["idDokumen"]))
{
List_idDoc.Add("'" + result.Rows[i]["idDokumen"] + "'".ToString());
}
else
{
}
}
arrayDoc = List_idDoc.ToArray();
return(arrayDoc);
}
public double cekKetersediaan_Term(string kata)
{
koneksi con = new koneksi();
con.openConnection();
string query = "SELECT \"Term\" FROM public.\"Term\" where \"Term\"='" + kata + "';";
DataTable result = con.getResult(query);
con.closeConnection();
if (result.Rows.Count >= 1)
{
con.openConnection();
double getIDF_Term = 0;
string QgetDF_Term = "SELECT \"IDF\" FROM public.\"Term\" where \"Term\"='" + kata + "';";
DataTable result2 = con.getResult(QgetDF_Term);
getIDF_Term = Convert.ToDouble(result2.Rows[0]["IDF"]);
con.stopAccess();
return(getIDF_Term);
}
else
{
return(0);
}
}
public void getDoc_MengandungQuery(string query)
{
string queryPakek = cariID_Term(query);
double[] hasilPembilang_Perkata;
double[] hasilPenyebut_Perkata;
if (queryPakek.Equals("0"))
{
Console.WriteLine("Tidak ada dokument yang memakai kata '" + query + "'");
Console.WriteLine("======================================================================");
}
else
{
koneksi con = new koneksi();
con.openConnection();
string queryDF = "SELECT \"idDokumen\", \"id_Term\", \"TF-IDF\" FROM public.\"bobotTerm\" where" +
" \"id_Term\"='" + queryPakek + "'";
DataTable result = con.getResult(queryDF);
string[] dataID_Term = new string[result.Rows.Count];
hasilTF_IDF_Perkata_Pembagi = new double[dataID_Term.Length];
hasilTF_IDF_Perkata_Pembilang = new double[dataID_Term.Length];
string[] dataTF_IDF_Term = new string[result.Rows.Count];
for (int i = 0; i < result.Rows.Count; i++)
{
int count = 1;
dataTF_IDF_Term[i] = result.Rows[i]["idDokumen"].ToString();
hasilTF_IDF_Perkata_Pembilang[i] = Convert.ToDouble(result.Rows[i]["TF-IDF"]);
hasilTF_IDF_Perkata_Pembagi[i] = Convert.ToDouble(result.Rows[i]["TF-IDF"]);
hasilPembilang_Perkata = new double[result.Rows.Count];
hasilPenyebut_Perkata = new double[result.Rows.Count];
if ((result.Rows.Count - 1) == i)
{
Console.WriteLine("dokument yang memakai kata '" + query + "' sebanyak " + result.Rows.Count + " dokumen yaitu " + dataID_Term[i]);
Console.WriteLine("======================================================================");
double TF_IDFperKata = cariTF_IDF(dataID_Term[i], queryPakek);
for (int w = 0; w < result.Rows.Count; w++)
{
string tempId = dataTF_IDF_Term[w];
hasilPembilang_Perkata[w] = Math.Round(hasilTF_IDF_Perkata_Pembilang[w] * hasil_TF_IDF_Query, 4);
hasilPenyebut_Perkata[w] = Math.Round(Math.Sqrt(hasilTF_IDF_Perkata_Pembagi[w] * hasil_TF_IDF_Query), 4);
Console.WriteLine("Nilai pembilang id doc '" + tempId + " = " + hasilPembilang_Perkata[w]);
Console.WriteLine("Nilai pembagi id doc '" + tempId + " = " + hasilPenyebut_Perkata[w]);
Console.WriteLine("======================================================================");
double hasilTotal_Pembilang = +hasilPembilang_Perkata[w];
for (int fr = 0; fr < idDoc_Terlibat.Length; fr++)
{
idList.Add(tempId);
if (result.Rows.Count - 1 == w)
{
idList_Array = idList.ToArray();
}
}
}
}
}
con.stopAccess();
}
}
public string[] runStemming_Tala(string text)
{
char[] delimiterChars = { ' ', ',', '.', ':', '\t', '-', '(', ')', '"', '`' };
System.Console.WriteLine($"Original text: '{text}'");
string[] words = text.Split(delimiterChars);
System.Console.WriteLine($"{text.Length} words in text:");
var count = new stemmingTala();
int getData = count.getCOunt();
int range = words.Length;
arrayWord = new string[range];
//kumpulan aturan tala Stemming//
string[] test_1 = { "lah", "kah", "pun" };
string[] test_2 = { "nya", "ku", "mu" };
string[] test_3 = { "meng", "menya", "menyi", "menyu", "menye", "menyo", "meny", "men", "mema", "memi", "memu", "meme", "memo",
"memo", "me", "peng", "penya", "penyi", "penyu", "penye", "penyo", "peny", "pen", "pema", "pemi", "pemu","peme","pemo", "pem", "di", "ter", "ke" };
string[] test_3replace = { "", "s", "s", "s", "s", "s", "s", "", "p", "p", "p", "p", "p", "", "", "", "s", "s", "s", "s", "s", "s", "",
"p", "p", "p", "p", "p", "", "", "", "" };
string[,] test_22 = { { "meng", "" }, { "menya", "s" }, { "menyi", "s" }, { "menyu", "s" }, { "menye", "s" }
,{ "menyo", "s" }, { "meny", "s" }, { "men", "s" }, { "mema", "p" }, { "memi", "p" }, { "memu", "p" }, { "meme", "p" }
,{ "memo", "p" }, { "mem", "" }, { "me", "" }, { "peng", "" }, { "penya", "s" }, { "penyi", "s" }, { "penye", "s" }, { "penyo", "s" }
,{ "peny", "s" }, { "pen", "" }, { "pema", "s" }, { "pemi", "s" }, { "pemu", "s" }, { "peme", "s" }, { "pemo", "s" }, { "pem", "" }
,{ "di", "" }, { "ter", "" }, { "ke", "" } };
string[,] test_3a = { { "ber", "" }, { "bel", "" }, { "be", "" }, { "per", "" }, { "pel", "" }, { "pe", "" } };
string[,] test_4das = { { "kan", "" }, { "an", "" }, { "i", "" } };
//proses memasukkan kata kedalam array//
for (int i = 0; i < words.Length; i++)
{
arrayWord[i] = text.Split(delimiterChars)[i].ToString().ToLower();
Console.WriteLine("[" + arrayWord[i] + "]");
//Proses filtering, atau menghilangkan kata sambung//
if (count.getKataSambung(arrayWord[i]) == true)
{
//proses menghilangkan wordList//
Console.WriteLine("[" + arrayWord[i] + "]" + "=> masuk kata sambung, kata dalam index ini dihapus");
arrayWord[i] = "";
}
//memeriksa apakah sudah kata dasar//
else if (count.getKataDasar(arrayWord[i]) == true)
{
arrayWord[i] = arrayWord[i];
this.con = new koneksi();
Console.WriteLine("[" + arrayWord[i] + "]" + "[ini kata dasar_ so Fix]");
}
//proses tala ke-1// hilangkan sandang kalau ada
else if (count.cekSandang(arrayWord[i]) > 0)//cek apakah punyak sandang
{
Console.WriteLine("[" + arrayWord[i] + "]" + " Masuk Proses Tala 1");
arrayWord[i] = count.replace_Sandang(arrayWord[i]);//proses hapus sandang
if (count.getKataDasar(arrayWord[i]) == true)
{
Console.WriteLine("sudah kata dasar jadi " + "[" + arrayWord[i] + "]");
}
else if (count.cekAkhiran(arrayWord[i]) > 0)//cek akhirannya
{
Console.WriteLine("[" + arrayWord[i] + "]" + " Masuk Proses Tala 1");
arrayWord[i] = count.replace_Akhiran(arrayWord[i]);//proses hapus akhiran
if (count.getKataDasar(arrayWord[i]) == true)
{
Console.WriteLine("sudah kata dasar jadi " + "[" + arrayWord[i] + "]");
}
else
{
Console.WriteLine("belum kata dasar " + "[" + arrayWord[i] + "]" + " Lanjut ke proses selanjutnya");
}
}
else
{
Console.WriteLine("belum kata dasar " + "[" + arrayWord[i] + "]" + " Lanjut ke proses selanjutnya");
if (count.cekAkhiran(arrayWord[i]) > 0) // cek apakah punyak akhiran
{
arrayWord[i] = count.replace_Akhiran(arrayWord[i]); //proses hapus akhiran
Console.WriteLine("diproses 2 jadi " + "[" + arrayWord[i] + "]");
}
else
{
Console.WriteLine("[" + arrayWord[i] + "]" + "Tidak Punyak akhiran, maka dilakukan menghilangkan awalan 1");
arrayWord[i] = count.replace_Awalan1(arrayWord[i]);
//runVSM_DB(arrayWord[i], id, getFrekunsiKata());//lakukan proses kelola DB
Console.WriteLine("Tidak Punyak akhiran, maka dilakukan menghilangkan awalan 1 menjadi " + "[" + arrayWord[i] + "]");
}
}
}
else if (count.cekAkhiran(arrayWord[i]) > 0)//cek akhirannya
{
Console.WriteLine("[" + arrayWord[i] + "]" + " Masuk Proses Tala 1");
arrayWord[i] = count.replace_Akhiran(arrayWord[i]);//proses hapus akhiran
if (count.getKataDasar(arrayWord[i]) == true)
{
//runVSM_DB(arrayWord[i], id, getFrekunsiKata());//lakukan proses kelola DB
Console.WriteLine("sudah kata dasar jadi " + "[" + arrayWord[i] + "]");
}
else
{
Console.WriteLine("belum kata dasar " + "[" + arrayWord[i] + "]" + " Lanjut ke proses selanjutnya");
if (count.getKataDasar(arrayWord[i]) == true)
{
Console.WriteLine("sudah kata dasar jadi " + "[" + arrayWord[i] + "]");
}
else if (count.cekawalan1(arrayWord[i]) > 0)//cek punyak awalan kah ???
{
Console.WriteLine("[" + arrayWord[i] + "]" + "Tidak Punyak akhiran, maka dilakukan menghilangkan awalan 1");
arrayWord[i] = count.replace_Awalan1(arrayWord[i]);
//runVSM_DB(arrayWord[i], id, getFrekunsiKata());//lakukan proses kelola DB
Console.WriteLine("Tidak Punyak akhiran, maka dilakukan menghilangkan awalan 1 menjadi " + "[" + arrayWord[i] + "]");
}
}
}
else if (count.cekawalan1(arrayWord[i]) > 0)
{
Console.WriteLine("[" + arrayWord[i] + "]" + "Tidak Punyak akhiran, maka dilakukan menghilangkan awalan");
arrayWord[i] = count.replace_Awalan1(arrayWord[i]);
//runVSM_DB(arrayWord[i], id, getFrekunsiKata());//lakukan proses kelola DB
Console.WriteLine("Tidak Punyak akhiran, maka dilakukan menghilangkan awalan 1 menjadi " + "[" + arrayWord[i] + "]");
}
else if (arrayWord[i].Length < 3)
{
arrayWord[i] = "";
}
else
{
arrayWord[i] = arrayWord[i];
//runVSM_DB(arrayWord[i], id, getFrekunsiKata());//lakukan proses kelola DB
Console.WriteLine("[" + arrayWord[i] + "]");
}
con.stopAccess();
array[i] = arrayWord[i];
}
return(array);
}