/**
* @see IDSequence::increaseId()
*
* @param int $inc
* @return int
*/
public function increaseId($inc = 1)
{
try {
$this->dbhelper->startTransaction();
$sql = "select current_value from id_sequence where id_key='%s' for update";
$current_value = $this->dbhelper->getOne($sql, $this->table_name);
if (empty($current_value)) {
$sql = "insert into id_sequence (id_key,current_value)values('%s',%d)";
$this->dbhelper->execute($sql, $this->table_name, $inc);
$next_id = $inc;
} else {
$sql = "update id_sequence set current_value = current_value + %d where id_key = '%s'";
$this->dbhelper->execute($sql, $inc, $this->table_name);
$next_id = $current_value + $inc;
}
$this->dbhelper->commit();
} catch (Exception $e) {
$this->dbhelper->rollback();
throw $e;
}
return $next_id;
}
/**
* Dokonuje bezpo¶redniej aktualizacji zbioru cech.
*
* @param array $comment Tablica asocjacyjna (wyraz => liczba wyst±pieñ).
* @param bool $positive Zbiór cech pozytywnych lub negatywnych?
*/
protected function updateVector(array $comment, $positive)
{
if ($positive) {
$pn = 'p';
} else {
$pn = 'n';
}
// uaktualnij i pobierz Nk
$sql = "UPDATE `datavect_{$this->idc}-4o` SET Nk{$pn} = Nk{$pn}+1";
$this->dbconn->query($sql);
$sql = "SELECT Nkp, Nkn FROM `datavect_{$this->idc}-4o`";
list($Nkp, $Nkn) = DBHelper::getRow($this->dbconn, $sql);
if ($positive) {
$logNk = log($Nkp + 1, 2);
} else {
$logNk = log($Nkn + 1, 2);
}
$NkSum = $Nkp + $Nkn;
$oneDivLogNK = 1 / $logNk;
// aktualizacja istniej±cych dfikp i dodanie nowych wyrazów
$sql = "INSERT INTO `datavect_{$this->idc}-4f` (word,dfikp,dfikn,WCikp,WCikn,CCi,idf) VALUES ";
$sql .= "(?, ";
$sql .= $positive ? "1, 0, ?, 0" : "0, 1, 0, ?";
$sql .= ", 2, 1)";
$sql .= "ON DUPLICATE KEY UPDATE dfik{$pn} = dfik{$pn} + 1";
$stmt = $this->dbconn->prepare($sql);
$stmt->bind_param("sd", $k, $oneDivLogNK);
foreach ($comment as $k => $v) {
$stmt->execute();
}
$stmt->close();
// uaktualnij zmienne statystyczne dla wszystkich wyrazów
$sql = "UPDATE `datavect_{$this->idc}-4f`\n SET WCik{$pn} = LOG2(dfik{$pn}+1)/{$logNk},\n CCi = 2 * LOG2((2 * GREATEST(dfikp,dfikn)) / (dfikp + dfikn)),\n idf = LOG({$NkSum}/(dfikp+dfikn));";
$this->dbconn->query($sql);
// pobierz wyliczone informacje statystyczne dla dodawanego komentarza
$comment_words = array();
$dbcom = "('" . implode("','", array_keys($comment)) . "')";
$sql = "SELECT word, CCi, idf FROM `datavect_{$this->idc}-4f` WHERE word IN {$dbcom}";
$res = $this->dbconn->query($sql);
while ($row = $res->fetch_row()) {
list($word, $CCi, $idf) = $row;
$comment_words[$word]['idf'] = $idf;
$comment_words[$word]['CCi'] = $CCi;
}
$res->free();
// Na podstawie zebranych danych wylicz wagi i znormalizuj warto¶ci dokum.
$sum = 0.0;
foreach ($comment as $k => &$v) {
$v = log($v + 0.5) * $comment_words[$k]['idf'] * $comment_words[$k]['CCi'];
if ($v == 0.0) {
$v = log($v + 0.5) * $comment_words[$k]['idf'] * 0.01;
}
$sum += $v;
}
if ($sum == 0.0) {
$sum = 1.0;
}
$mul = 1 / $sum;
foreach ($comment as $k => &$v) {
$v *= $mul;
}
unset($comment_words);
// dodaj nowy komentarz
$sql = "INSERT INTO `datavect_{$this->idc}-4s` (pn, last_act) VALUES ('{$pn}', 0)";
$this->dbconn->query($sql);
$sql = "INSERT INTO `datavect_{$this->idc}-4p` (subcat, word, weight) VALUES ";
$subcat = $this->dbconn->insert_id;
foreach ($comment as $k => $v) {
$sql .= "({$subcat}, '{$k}', {$v}),";
}
$sql[strlen($sql) - 1] = ' ';
$this->dbconn->query($sql);
// usuñ najdawniej nieu¿ywan± grupê, je¶li liczba grup
// jest wiêksza od $this->maxComments
$sql = "SELECT COUNT(subcat) FROM `datavect_{$this->idc}-4s`";
$count = DBHelper::getOne($this->dbconn, $sql);
if ($count <= $this->maxComments) {
return;
}
// wybierz najdawniej nieu¿ywan± grupê
$sql = "SELECT subcat FROM `datavect_{$this->idc}-4s` ORDER BY last_act DESC LIMIT 1";
$subcat = DBHelper::getOne($this->dbconn, $sql);
// usuñ z³±czane grupy, które wchodz± w sk³ad nowej grupy g³ównej
$sql = "DELETE `datavect_{$this->idc}-4s` WHERE subcat = {$subcat}";
$this->dbconn->query($sql);
$sql = "DELETE `datavect_{$this->idc}-4p` WHERE subcat = {$subcat}";
$this->dbconn->query($sql);
}
/**
* Dokonuje bezpo¶redniej aktualizacji zbioru cech.
*
* @param array $comment Tablica asocjacyjna (wyraz => waga po log.).
* @param bool $positive Zbiór cech pozytywnych lub negatywnych?
*/
protected function updateVector(array $comment, $positive)
{
if ($positive) {
$pn = 'p';
} else {
$pn = 'n';
}
// uaktualnij i pobierz Nk
$sql = "UPDATE `datavect_{$this->idc}-2o` SET Nk{$pn} = Nk{$pn}+1";
$this->dbconn->query($sql);
$sql = "SELECT Nk{$pn} FROM `datavect_{$this->idc}-2o`";
$Nk = DBHelper::getOne($this->dbconn, $sql);
$logNk = log($Nk + 1, 2);
$oneDivLogNK = 1 / $logNk;
// aktualizacja istniej±cych i dodanie nowych
$sql = "INSERT INTO `datavect_{$this->idc}-2` (word,dfikp,dfikn,WCikp,WCikn,CCi,widp,widn) VALUES ";
$sql .= "(?, ";
$sql .= $positive ? "1, 0, ?, 0" : "0, 1, 0, ?";
$sql .= ", LOG2(dfik{$pn}), ";
$sql .= $positive ? "?, 0" : "0, ?";
$sql .= ") ";
$sql .= "ON DUPLICATE KEY UPDATE dfik{$pn} = dfik{$pn} + 1, wid{$pn} = wid{$pn} + ?";
$stmt = $this->dbconn->prepare($sql);
$stmt->bind_param("sddd", $k, $oneDivLogNK, $v, $v);
foreach ($comment as $k => $v) {
$stmt->execute();
}
$stmt->close();
// Aktualizacja elementów globalnych (dotycz±cych wszystkich wyrazów).
$sql = "UPDATE `datavect_{$this->idc}-2` SET WCik{$pn} = LOG2(dfik{$pn}+1)/{$logNk}, CCi = LOG2((2 * GREATEST(dfikp,dfikn)) / (dfikp + dfikn))";
$this->dbconn->query($sql);
}
