tfidf

<?php
require("konektor.php");
//GET STOPLIST DATA
$sql="SELECT stoplist from tb_stoplist";
$result3 = mysql_query ($sql) or die ("Kesalahan pada perintah SQL!");
while($row=mysql_fetch_object($result3))
{
  $db_stoplist[]=$row->stoplist;
}
//AWAL TOKENIZING
$keyword=$this->key;
$tokenizing=explode(' ',$keyword);
$ct=count($tokenizing);
?>
<strong>Keyword: </strong><?php echo $keyword;?>
<br>
<strong>Panjang String: </strong><?php echo ($ps-2);?>
<?php
//AKHIR TOKENIZING
//FILTERING
$filtering=array_diff($tokenizing,$db_stoplist);
//ANALYZING
//Ambil Jumlah Dokumen
$D=0;
$sqldok = "SELECT terjemah FROM tb_hadits ORDER BY id_hadits ASC";
$result2 = mysql_query ($sqldok) or die ("Kesalahan pada perintah SQL!");
$dokumen=array();
while($row2 = mysql_fetch_array($result2))
{
  ++$D;
  //Ambil Kata Tiap Dokumen, Dilakukan Tokenizing,Filtering,Penghilangan Duplikasi Token
  //TOKENIZING DOKUMEN
  $replace=preg_replace('/[":;.,!?]/', ' ',$row2['terjemah']);
  $replace=preg_replace('/\s+/', ' ', $replace);
  $token_dok=explode(' ',trim($replace);
  $ct=count($token_dok);
  //FILTERING DOKUMEN
  $temp_dok=array_diff($token_dok,$db_stoplist);
  $dokumen=array_merge($dokumen,$temp_dok);
}
//Penghilangan Duplikasi Token
$token=array_unique($dokumen);
$jml_token=count($token);
echo("<center><strong><-- HASIL PENCARIAN --></strong></center>");
//Area TF-IDF
$jumlahkosong=0;
$counttf=0;
$countw=0;
$df=0;
for($i=0;$i<$jml_token;$i++)
{
  //Area TF
  for($j=0;$j<=$D+1;$j++)
  {
    if($j==0)
    {
      //cek kata kunci(kk)
      $adakunci=0;
      $kunci=$token[$i];
      for($k=0;$k<count($filtering);$k++)
      {
      $katakey=$filtering[$k];
        if($kunci==$katakey)
        {
          $adakunci=1;
        }
      }
      $tf[$i][$j]=$adakunci;
      if($adakunci==0)
      {
        $jumlahkosong++;
      }
    }
    else if($j==($D+1))
    {
      $hasilidf=$D/$df;
      $hasiltemp=log10($hasilidf);
      $tf[$i][$j]=$hasiltemp;
    }
    else
    {
      //cek di dokumen
      $kunci=$token[$i];
      $urutandok=0;
      $akdok=0;
      $nodok=$j;
      $sqldok = "SELECT * FROM tb_hadits ORDER BY id_hadits ASC";
      $result4 = mysql_query ($sqldok) or die ("Kesalahan pada perintah SQL!");
      while($row4 = mysql_fetch_array($result4))
      {
        if($urutandok==$nodok-1)
        {
          //Tokenizing Dokumen
          $isidokumen=preg_replace('/[":;.,!?]/', '',$row4['terjemah'])
          $isidokumen=strtolower($isidokumen);
          $isidokumen=$isidokumen."#e";
          $ps2=strlen($isidokumen);
          $y=1;
          $x=0;
          $ct=0;
          unset($tokendok2);
          for ($ii=0;$ii<$ps2;$ii++)
          {
            if(substr($isidokumen,$ii,1)==" " || substr($isidokumen,$ii,2)==substr($isidokumen,$ps2-2,2))
            {
              $tempkata=substr($isidokumen,$x,$y-1);
              $tokendok2[++$ct]=$tempkata;
              $x=$ii+1;
              $y=0;
            }
            $y=$y+1;
          }
          //Filtering Dokumen
          $jmlada=0;
          $cnfilter=0;
          unset($filterdok2);
          for($ii=0;$ii<count($tokendok2);$ii++)
          {
            $katatoken=$tokendok2[$ii];
            $sqlfil = "SELECT stoplist from tb_stoplist";
            $result5 = mysql_query ($sqlfil) or die ("Kesalahan pada perintah SQL!");
            while($row5 = mysql_fetch_array($result5))
            {
              $katafilter=$row5['stoplist'];
              if($katatoken==$katafilter){
                $jmlada=$jmlada+1;
              }
            }
            if($jmlada==0)
            {
              $filterdok2[$cnfilter]=$katatoken;
              $cnfilter=$cnfilter+1;
            }
            $jmlada=0;
          }
          for ($kk=0;$kk<count($filterdok2);$kk++)
          {
          $hsfilter=$filterdok2[$kk];
            if($kunci==$hsfilter){
              $akdok=$akdok+1;
            }
          }
        }
        $urutandok++;
      }
      $tf[$i][$j]=$akdok;
    }
    $tnilaidf=$tf[$i][$j];
    if($tnilaidf>=1 && $j!=0)
    {
      $df=$df+1;
    }
  }
  $df=0;
  //Akhir Area TF
  //Area W
  for ($k=0;$k<=$D;$k++)
  {
    $w[$i][$k]=$tf[$i][$k]*$tf[$i][($D+1)];
  }
  //Akhir Area W
}
//Awal Area Pehitungan Algoritma TF-IDF
for($i=0;$i<$jml_token;$i++)
{
  for($j=0;$j<=$D;$j++)
  {
    $temptf=$tf[$i][0];
    if($temptf=="1")
    {
      $tfidf[$j]=$tfidf[$j]+$w[$i][$j];
    }
  }
}
//Akhir Area Perhitungan Algoritma TF-IDF
//NB:
//Area Sorting Nilai tfidf menggunakan Algoritma Bubble Sort
//array tfidf ada kk ke 0, jadi dimulai dari index 1
for($j=0;$j<=$D;$j++){
  $tfidfno[$j]="Dokumen ".$j;
}
for($j=1;$j<=$D;$j++)
{
  for($jj=1;$jj<$D;$jj++)
  {
  $tempj=$tfidf[$j];
  $tempjj=$tfidf[$jj];
    if($tempj>=$tempjj)
    {
      $tj=$tfidf[$jj+1];
      $tjn=$tfidfno[$jj+1];
      $tfidf[($jj+1)]=$tfidf[$jj];
      $tfidfno[($jj+1)]=$tfidfno[$jj];
      $tfidf[$jj]=$tj;
      $tfidfno[$jj]=$tjn;
    }
  }
}
//Akhir Area TF-IDF
//Area VSM
//hitung transformasi w menjadi kuadrat dan sqrt total dari tiap dokumen
for($i=0;$i<=$jml_token;$i++)
{
  if($i==$jml_token)
  {//menghitung sum akar/sqrt tiap dokumen+kk
    for($j=0;$j<=$D;$j++)
    {
      $tvsmtot=0;
      for($k=0;$k<$jml_token;$k++)
      {
        $tnvsm=$vsmw[$k][$j];
        $tvsmtot=$tvsmtot+$tnvsm;
      }
      $tvsmtot=sqrt($tvsmtot);
      $vsmw[$i][$j]=$tvsmtot;
    }
  }
  else
  {//menghitung kuadrat tiap token dalam dokumen(kuadrat w)
    for($l=0;$l<=$D;$l++)
    {
      $tvsm=$w[$i][$l];
      $tvsm=$tvsm*$tvsm;
      $vsmw[$i][$l]=$tvsm;
    }
  }
}
//akhir area hitung transformasi w menjadi kuadrat dan sqrt total dari tiap dokumen
//hitung kk*D memanfaatkan vsmw index ke 0(kk) * D tiap dokumen
for($i=0;$i<=$jml_token;$i++)
{
  if($i==$jml_token)
  {//untuk menghitung sum kk dot D tiap dokumen
    for($j=0;$j<=$D;$j++)
    {//index ke 0 (kk*kk) dimasukan tapi tidak akan dipakai
      $tvsmkk=0;
      for($k=0;$k<$jml_token;$k++)
      {
        $tvsmkkd=$vsmkk[$k][$j];
        $tvsmkk=$tvsmkk+$tvsmkkd;
      }
      $vsmkk[$i][$j]=$tvsmkk;
    }
  }
  else
  {//untuk menghitung kk*D tiap dokumen
    for($l=0;$l<=$D;$l++)
    {
      if($l>=1)
      {
        $temptf=$tf[$i][$l];
        $tempvsmkk=$vsmw[$i][0];
        $tempvsmkk2=$temptf*$tempvsmkk;
        $vsmkk[$i][$l]=$tempvsmkk2;
      }
    }
  }
}
//perhitungan penilaian cosine setiap dokumen dengan vsm
for($j=0;$j<=$D;$j++){
  $sumkkd=$vsmkk[$jml_token][$j];
  $sqrtkk=$vsmw[$jml_token][0];
  $sqrtd=$vsmw[$jml_token][$j];
  @$tcosine=$sumkkd/($sqrtkk*$sqrtd);
  $cosine[$j]=$tcosine;
}
for($j=0;$j<=$D;$j++)
{
  $cosineno[$j]=$j;
}
//NB:
//Area Sorting Nilai cosine vsm menggunakan Algoritma Bubble Sort
//array vsm ada ke 0(tidak dipakai), jadi dimulai dari index 1
for($j=1;$j<=$D;$j++)
{
  for($jj=1;$jj<$D;$jj++)
  {
    $tempj=$cosine[$j];
    $tempjj=$cosine[$jj];
    if($tempj>=$tempjj)
    {
      $tj=$cosine[$jj+1];
      $tjn=$cosineno[$jj+1];
      $cosine[($jj+1)]=$cosine[$jj];
      $cosineno[($jj+1)]=$cosineno[$jj];
      $cosine[$jj]=$tj;
      $cosineno[$jj]=$tjn;
    }
  }
}
//Akhir Area VSM
//Tampil Hadist
$sqldok = "SELECT
             b.id_bab,
             b.bab,
             h.id_hadits,
             h.judul,
             h.hadits,
             h.terjemah,
             h.no_hadits
           FROM
             tb_bab b,
             JOIN tb_hadits h USING(id_bab)
           ORDER BY h.id_hadits ASC";
$result7 = mysql_query($sqldok) or die ("Kesalahan pada perintah SQL!");
$no=0;
while($row7 = mysql_fetch_array($result7))
{
  $bab[$no]=$row7['bab'];
  $judul[$no]=$row7['judul'];
  $nomor[$no]=$row7['no_hadits'];
  $hadits[$no]=$row7['hadits'];
  $terjemah[$no]=$row7['terjemah'];
  $no=$no+1;
}
echo("<br>");
for($i=0;$i<count($cosine);$i++)
{
  $tmpcosine=$cosine[$i];
  $tmpcosineno=$cosineno[$i];
  $tmpcosineno=$tmpcosineno-1;
  if($tmpcosine>0){
echo <<<HASIL
<table>
<tr>
<td style="padding:5px;">
Score : <strong>$tmpcosine</strong>
</td>
</tr>
<tr>
<td style="padding:5px;">
Bab Hadits: <strong>{$bab[$tmpcosineno]}</strong> No. <strong>{$nomor[$tmpcosineno]}</strong>. Judul: <strong>{$judul[$tmpcosineno]}</strong>
</td></tr>
<tr>
<td align="right" style="padding:5px; font-size:12pt;">
{$hadits[$tmpcosineno]}
</td>
</tr>
<tr><td style="padding:5px;">
{nl2br($terjemah[$tmpcosineno])}
</td></tr>
</table>
HASIL;
  }
}
?>