Lingvistica's Extract Semantica Data

200000000) { echo "

The selected file ".$text_file." with the size ".$file_size." is larger than the 200MB we can process on-line. Call us on +31-4807 8750 for advice

"; exit; } if ($invr_lgrm == null) { $invr_lgrm=0; } $file_time=$file_size/500000; $file_time=round($file_time, 2); $time_text=$file_time." secondes"; if ($file_time > 60) { $file_time=$file_time/60; $file_time=round($file_time, 2); $time_text=$file_time." minuten"; } // echo "
INVR FILE ".$invr_file."FILETIME ".$file_time." seconds
"; echo "

Bekijk het bestand ".$invr_file."
Het bestand is ".$file_size." bytes groot. De verwerkingstijd is ongeveer ".$time_text."

"; // echo "
INVR CORP ".$invr_corp." INVR NAME ".$invr_name." INVR bzv1 ".$invr_bzv1."INVR LGRM ".$invr_lgrm." INVR CHBT ".$invr_chbt." INVR FILE ".$invr_file." INVR DOMN ".$invr_domn; // echo "
De taal is ".$invr_taal; $userip = $_SERVER["REMOTE_ADDR"]; $userip=$_SERVER['REMOTE_ADDR']; if ($userip != "84.24.147.72") { echo "This action cannot be completed for this IP Address. Please contact us"; exit; } $requester= $_SERVER["HTTP_REFERER"]; $invr_ipad=$userip; $file_dtls=explode("/",$invr_file); // print_r($file_dtls); $invr_corp=$file_dtls[2]; $text_file="https://www.semanta.nl/semanta".$invr_file; if ($invr_corp== null) { $invr_corp=$invr_domn; } echo "

Wat zegt Wikipedia over ".$invr_corp."

"; $date = gmdate("Y-m-d"); setlocale(LC_ALL, 'nl_NL'); $localtime = localtime(); $tijd=$localtime[2].":".$localtime[1].":".$localtime[0]; $date = gmdate("Y-m-d")." ".$tijd; echo $date; $smta_rslt=""; $sw=0; /////////////////////// $file_dtls=explode("/",$invr_file); // print_r($file_dtls); $file_corp=$file_dtls[1]; $file_text=$file_dtls[2]; $text_itms=explode(".",$file_text); $file_corp=$text_itms[0]; $file_name=$file_dtls[3]; $invr_corp=$file_corp; $userip=$_SERVER['REMOTE_ADDR']; ?> ".$invr_cont.""; ////////////////////// BUILD IPAD TABLE AND COUNTERS //////////// $base_rslt=xtrt_xxxx($invr_cont,$invr_taal,"base"); print_r($base_rslt); $sql0 = "SELECT * FROM base where `rchs`= '$invr_corp' or `lnks` = '$invr_corp' order by `rchs` asc"; mysql_query("SET CHARACTER SET utf8"); mysql_query("SET NAMES utf8"); $tablresult = mysql_query($sql0, $link); $tabl_num = mysql_num_rows($tablresult); $corp_num=$tabl_num; // echo "
Aantallen ".$sql0." ".$tabl_num." SQL "; while ($row = mysql_fetch_assoc($tablresult)) { $tabl_ipad="LINGVISTICA"; $id=$row['id']; $ipad=$row['ipad']; $ipad_array[$ipad]=$ipad_array[$ipad]+1; $domn=$row['domn']; $rule_domn=$domn; $domn_array[$domn]=$domn_array[$domn]+1; $lnks=$row['lnks']; $lnks_array[$lnks]=$lnks_array[$lnks]+1; $rchs=$row['rchs']; $rchs_array[$rchs]=$rchs_array[$rchs]+1; $taal=$row['taal']; if ($invr_taal == null) { $invr_taal=$taal; } $levl=$row['levl']; $datm=$row['datm']; $taal_array[$taal]=$taal_array[$taal]+1; $tlnks=$tlnks+$lnks; $trchs=$trchs+$rchs; $tlevl=$tlevl+$levl; $file_find=""; } // print_r($domn_array); if (count($domn_array) > 0) { echo ""; echo ""; foreach($domn_array as $domn_key => $domn_val) { echo ""; } echo "

"; } ///////////////////////////////////////////////////////////////////////////////////////////// $file_dtls=explode("/",$invr_file); // print_r($file_dtls); $file_corp=$file_dtls[1]; $file_text=$file_dtls[2]; $text_itms=explode(".",$file_text); // print_r($text_itms); $file_corp=$text_itms[0]; $file_name=$text_itms[1]; if ($file_name != "data") { // echo "
TCH MAAR AANNDEERRSS ".$file_name."CORP FILE ".$file_corp." FILE TEXT ".$file_text."
"; $file_array[0]=$invr_file; echo "".$file_val.""; } if ($handle = opendir($invr_file)) { { while (false !== ($entry = readdir($handle)) ) { // echo "
ENTRY ".$entry; $file_array[$e]= $entry; $e=$e+1; } } } asort($file_array); // print_r($file_array); echo "
AANTAL TEKSTBESTANDEN ".count($file_array); if (count($file_array) == 1) { $file_array[0]=$invr_file; } foreach($file_array as $file_key => $file_val) { // echo "
INVR CONT ".$invr_cont."FILE VAL ".$file_val; $file_dtls=explode(".",$file_val); $file_drct=$file_dtls[0]; $file_xtnt=$file_dtls[1]; // echo "
PER FILE INVR CONT ".$invr_cont."FILE VAL ".$file_val." FILE XTNT ".$file_xtnt; // echo "".$file_val.""; $file_size=filesize($text_file); $open_corp=$text_file; $text_file=$invr_corp."/".$file_val; echo ""; // echo "
PER FILE TEXT FILE ".$file_val. $f = fopen($file_val, "r"); $word_count=0; $line_count=0; $full_docm=""; while ($line = fgets($f, 1000)) { $line=trim($line); $line=strtolower($line); $line=str_replace(" - investeerders gezocht? nederlands grootste investeerders netwerk.","",$line); //////// $uline=urlencode($line); $uline=str_replace("%28%F0%91%F0%B5%F0%BB%F0%B0%F1%80%F1%83%F1%81%F1%8C%29","",$uline); $uline=str_replace("%F0%A0%F1%83%F1%81%F1%81%F0%BA%F0%B8%F0%B9","",$uline); $uline=str_replace("%28%F0%9A%F0%B0%F0%B7%F0%B0%F1%85%F1%81%F1%82%F0%B0%F0%BD%29","",$uline); $uline=str_replace("%28%F0%A0%F0%BE%F1%81%F1%81%F0%B8%F1%8F%29","",$uline); $uline=str_replace("%28%F8%A7%F9%84%F9%85%F9%85%F9%84%F9%83%F8%A9","",$uline); $uline=str_replace("%28%F0%A3%F0%BA%F1%80%F0%B0%F1%97%F0%BD%F0%B0%29","",$uline); $uline=str_replace("%F1%83%F0%BA%F1%80%F0%B0%F1%97%F0%BD%F1%81%F1%8C%F0%BA%F0%B0","",$uline); $uline=str_replace("%E0%B9%84%E0%B8%97%E0%B8%A2","",$uline); $uline=str_replace("%F8%A7%F9%84%F8%B9%F8%B1%F8%A8%F9%8A%F9%87","",$uline); $uline=str_replace("%F8%A7%F9%84%F8%B3%F8%B9%F9%88%F8%AF%F9%8A%F8%A9%29","",$uline); $uline=str_replace("%28%F0%91%F1%8A%F0%BB%F0%B3%F0%B0%F1%80%F0%B8%F1%8F%29","",$uline); $uline=str_replace("%F0%B1%F1%8A%F0%BB%F0%B3%F0%B0%F1%80%F1%81%F0%BA%F0%B8","",$uline); $uline=str_replace("%F8%A7%F9%84%F8%B9%F8%B1%F8%A8%F9%8A%F8%A9","",$uline); $uline=str_replace("%28%EE%95%EE%BB%EE%BB%EE%AC%EE%B4%EE%B1%29","",$uline); $uline=str_replace("%EE%95%EE%BB%EE%BB%EE%B7%EE%BD%EE%B9%EE%BA%EE%AC","",$uline); $uline=str_replace("%F0%B1%F0%B5%F0%BB%F0%B0%F1%80%F1%83%F1%81%F0%BA%F0%B0%F1%8F","",$uline); $uline=str_replace("%BF%BF","",$uline); $uline=str_replace("%E2%A9"," copyright ",$uline); $uline=str_replace("%E3%A8","e",$uline); $uline=str_replace("%E3%AB","e",$uline); $uline=str_replace("%E2%BB","",$uline); $uline=str_replace("%E3%A8","e",$uline); $uline=str_replace("%E3%A9","e",$uline); $uline=str_replace("%E3%AF","i",$uline); $uline=str_replace("%E2%B0"," Celsius",$uline); $uline=str_replace("%E3%B1","n",$uline); $uline=str_replace("%E3%A7","c",$uline); $uline=str_replace("%E3%BC","u",$uline); $uline=str_replace("%E3%BA","u",$uline); $uline=str_replace("%E3%AA","e",$uline); $uline=str_replace("%E3%A1","a",$uline); $uline=str_replace("%E3%A2","e",$uline); $uline=str_replace("%E3%A3","a",$uline); $uline=str_replace("%E3%A1","a",$uline); $uline=str_replace("%E5%A1","v",$uline); $uline=str_replace("%E4%8C","c",$uline); $uline=str_replace("%E3%96","o",$uline); $uline=str_replace("%E3%B3","o",$uline); $uline=str_replace("%E3%B6","oe",$uline); $uline=str_replace("%E2%AE","tm",$uline); $uline=str_replace("%A9","",$uline); $uline=str_replace("%B4","",$uline); $uline=str_replace("%E3","e",$uline); $uline=str_replace("%EB","e",$uline); $uline=str_replace("%EF","i",$uline); $uline=str_replace("%F6","o",$uline); // echo "
ULINE ".$uline; $line=urldecode($uline); $strt_line=substr($line,10,256); $line=str_replace(" - investeerders gezocht? nederlands grootste investeerders netwerk.","",$line); if ($srelt != $strt_line) { $srelt=$strt_line; //////// if (substr($line,0,10) == "the_title:") { // echo "
STRT LINE ".$strt_line; $titl_array[$strt_line]=$titl_array[$strt_line]+1; $nt=$nt+1; if ($nt > 0) { // echo ""; $title=substr($line,10,256); // echo ""; $sgmt_rslt=bepl_sgmt($title); // print_r($sgmt_rslt); foreach($sgmt_rslt as $sgmt_key => $sgmt_val) { if ($sgmt_key != null) { if ($krelt != $sgmt_key) { // echo "
<category><pattern>".$sgmt_key."</pattern><template>".$title."</template></category>"; // echo ""; $krelt=$sgmt_key; } } } // echo ""; $stitle=str_replace(" ","+",$title); // $ltitle="".$title.""; // echo "
".$ltitle; // $conv_line=conv_text($line); // print_r($conv_line); $titl_pntr=ftell($f); if ($titl_sw == 0) { $titl_sw=1; $domn_titl=$title; if ($userip != "84.24.147.72") { // echo "This action cannot be completed for this IP Address. Please contact us"; // break; } } if (substr($line,0,8) == "THE_URL:") { $url_dtls=explode("?",$line); $url=substr($line,8,$title_lngt); $aiml_url="<url>".$invr_file."</url></link>"; $line=str_replace(".","xprd",$line); if (count($url_dtls) > 1) { echo "
".$aiml_url; // echo "
URL Verfijning:".$url_dtls[1]; } } if ($line != "*" and substr($line,0,1) != "*" ) { $line=$line."."; } $line=str_replace("__","",$line); $line=str_replace("_"," ",$line); $line=str_replace("› ","",$line); $line=str_replace("|"," ",$line); $line=str_replace("&"," and ",$line); $line=str_replace("#"," ",$line); $line=str_replace("«"," ",$line); $line=str_replace("»"," ",$line); $line=str_replace('"','',$line); $line=str_replace("(BUTTON)","",$line); $line=str_replace(" ","",$line); $line=str_replace(" ","",$line); $line=str_replace("Lees meer",".lees meer.",$line); $line=str_replace("v. Chr","BC.",$line); $line=str_replace("V. Chr","BC.",$line); $line=str_replace("N. Chr","AC.",$line); $line=str_replace(" ","",$line); if ($line != null and substr($line,0,8) != "THE_URL:") { // echo ""; // echo ""; } $title_lngt=(strlen($line)-10); if (substr($line,0,1) == "*" ) { // $line=str_replace("goto","",$line); $line_ptrn=substr($line,0,256); // echo "
<button>".$line_ptrn."</button>"; // echo ""; } if ($line != null and $line != "*." and $line != ".") { $full_docm=$full_docm.$line; $full_star=explode("*",$line); // print_r($full_star); foreach($full_star as $star_key => $star_val) { if ($star_key != 0 or $star_val != null or $star_val != " ." or $star_val !="") { $star_val=str_replace(".","",$star_val); $star_val=str_replace("-","",$star_val); $star_val=str_replace(" "," ",$star_val); if ($star_val != null) { if ($srelt != $star_val and $star_val != null) { $srelt=$star_val; $star_val=trim($star_val); $star_array[$star_val]=$star_array[$star_val]+1; } } // echo ""; } } } } } } } } // print_r($star_array); arsort($titl_array); foreach($titl_array as $titl_key => $titl_val) { $mtitl_key=str_replace("the_title:overzicht van beroepen met ","",$titl_key); $mtitl_key=str_replace("| nationaleberoepengids.nl","",$mtitl_key); $mtitl_key=str_replace("the_title:","",$mtitl_key); $mtitl_key=str_replace(" - investeerders gezocht? nederlands grootste investeerders netwerk.","",$mtitl_key); // echo "
VAL ".$titl_val." KEY ".$mtitl_key; } if (count($star_array) > 0) { foreach($star_array as $star_key => $star_val) { $star_key=trim($star_key); // echo ""; // echo ""; } } //////////// echo "

TITLE-".$nt." ".$strt_line."
TITLE ".$title."
SEGMENTEN ".$sgmt_key."	".$sgmt_val."
MODIFIED ".substr($line,0,256)."
".urlencode($line)."	".$line."
<category><pattern>".$line_ptrn."</pattern><template>
STAR DOM".$star_val."
<button>".urlencode($star_key)."</button>
<button>".$star_key."</button>

"; $paiml_link=$aiml_text.$aiml_url; echo "
FULL DOCM ".strlen($full_docm); $full_lnes=explode(".",$full_docm); echo ""; // echo "

".$aiml_text." ".$aiml_url."

"; $splt_sgns="+,&"; // echo "
INVR BZV1 ".$invr_bzv1; asort($full_lnes); foreach($full_lnes as $full_key => $full_val) { $full_val=str_replace(",","",$full_val); $full_val=str_replace(" "," ",$full_val); $full_val=str_replace("-"," ",$full_val); $full_val=str_replace("_"," ",$full_val); $full_val=str_replace("&"," ",$full_val); $full_val=str_replace("]"," ",$full_val); $full_val=str_replace("["," ",$full_val); $full_val=str_replace("'"," ",$full_val); $full_val=str_replace(")","",$full_val); $full_val=str_replace("(","",$full_val); $full_val=str_replace(".png","",$full_val); $full_itms=explode(" ",$full_val); $ni=0; $itms_rslt=""; // print_r($full_itms); foreach($full_itms as $itms_key => $itms_val) { if (strlen($itms_val) > 3) { $gw27_rslt=xtrt_xxxx($itms_val,$invr_taal,"gw27",4); foreach($gw27_rslt as $gw27_key => $gsw27_val) { $gw27_dtls=explode("!!",$gw27_key); // print_r($gw27_dtls); $gw27_domn=$gw27_dtls[2]; $gw27_gram=$gw27_dtls[3]; $gw27_word=$gw27_dtls[4]; if ($gw27_domn == "grammatica") { if ($gw27_gram == "znw" or $gw27_gram == "mvd" or $gw27_gram == "wwd" or $gw27_gram == "prp" or substr($gw27_gram,1,2) == "nm" or substr($gw27_gram,1,2) == "pe") { // echo "
".$gw27_gram." ITMS_VAL ".$itms_val; echo "
".$full_val." ITMS_VAL ".$itms_val; $itms_array[$full_val."@".$itms_val]=$itms_array[$full_val."@".$itms_val]+1; } } } } } } print_r($itms_rslt); ?> Toon sleutelwooorden $full_val) { echo "
KEY ".$full_key." VAL ".$full_val.""; $full_val=str_replace("-"," ",$full_val); $full_val=str_replace("_"," ",$full_val); $full_val=str_replace("&"," ",$full_val); $full_val=str_replace("]"," ",$full_val); $full_val=str_replace("["," ",$full_val); $full_val=str_replace("'"," ",$full_val); $full_val=str_replace(")","",$full_val); $full_val=str_replace("(","",$full_val); $full_val=str_replace(".png","",$full_val); if ($full_val != null) { $tlkd_pos = strpos($full_val,$invr_cont); if ($tlkd_pos === false) { echo "
KEY ".$full_key." VAL ".$full_val.""; $desc_array["v@".$full_key]=$desc_array["v@".$full_key]+1; } else { if ($frelt != $full_val) { $sgmt_rslt=bepl_sgmt($full_val); foreach($sgmt_rslt as $itms_key => $itms_val) { if ($irelt != $itms_val) { $irelt=$itms_val; // echo "
ITMS VAL ".$itms_val." SPEC RSLT ".$spec_rslt; $spec_rslt=bepl_spec($itms_val); // echo "
SPEC RSLT ".$spec_rslt; // echo "
ITMS VAL ".$itms_val." SPEC RSLT ".$spec_rslt; echo ""; echo ""; $frelt=$full_val; } } } // echo ""; } } } // echo "
></template></category>"; // arsort($itms_array); echo "

"; // Meest frequente sleutelwoorden $ni=0; // print_r($itms_array); // echo "
".count($itms_array)." Sleutelwoorden in de volgorde van hun frequentie."; $ni_max=count($itms_array); if ($ni_max > 10) { $ni_max=$ni_max/5; $ni_max=round($ni_max, 0); // echo "
Ik laat er ".$ni_max." zien."; } echo "

"; // Alle sleutelwoorden $ni_max=count($itms_array); echo "

".$full_val." ".$spec_rslt."	<button><".$full_val."></button>
".$full_val."

"; echo ""; foreach($itms_array as $itms_key => $itms_val) { $itms_dtls=explode("@",$itms_key); $itms_word=$itms_dtls[0]; $itms_text=$itms_dtls[1]; $sgmt_keys=bepl_sgmt($itms_wrds); // echo ""; // print_r($sgmt_keys); foreach($sgmt_keys as $keys_key => $keys_val) { // echo ""; } if ($wrelt != $word_val and $word_val != $itms_text) { echo ""; $wrelt=$word_val; } } // print_r($word_itms); if ($ni < $ni_max) { if ($ri == 3) { // echo ""; } if ($ri== 3) { echo ""; $ri=0; } $litms_text="".$itms_text.""; // echo ""; $ri=$ri+1; $sgmt_rslt=bepl_sgmt($itms_word); $sgmt_text=""; foreach($sgmt_rslt as $sgmt_key => $sgmt_val) { $sgmt_text=$sgmt_key."
".$sgmt_text; } echo ""; $ni=$ni+1; } } echo "

TEXT ".$itms_text."
".$keys_key; } $word_itms=explode("*",$itms_word); foreach($word_itms as $word_key => $word_val) { if ($wrelt == $word_val) { echo "
	".$itms_text."
".$word_val."	".$itms_text."

".urlencode($itms_text)."	".$litms_text."	".$sgmt_text." ".$litms_text."