}
//string keys to numeric
$candidates = array_values($candidates);
//aggregate to size N+1
$current_generation = keyword_aggregate($candidates);
foreach ($current_generation as $group) {
$current_sets[] = $group[0];
//add result set to current sets
}
$round_count++;
}
echo '</table>';
//insert results to databse
$product = $row['url'];
foreach ($previous_sets as $keywordset) {
$keywordset_string = kwset_to_string($keywordset);
$query = "INSERT INTO keywordset_product(keyword_set, product) VALUES('{$keywordset_string}', '{$product}')";
$all_insert_attemps++;
$ret_val = mysql_query($query);
if ($ret_val) {
$insert_success++;
}
}
$previous_keywords = $current_keywords;
$previous_sets = $current_sets;
}
mysql_query("COMMIT");
echo "Insert success: {$insert_success} / {$all_insert_attemps}<br/>";
}
$end_time = microtime(true);
echo 'processed: ' . ($end_time - $start_time) . ' ms';
$occur = occurrence($keyword_set, $all_splitted_keyword_sets);
$kwset_occur_mapping[$kwset_string] = $occur;
}
//calculate the index here, finally!
$intersection_length = $kwset_occur_mapping[$kwset_string];
$suppset_occur1 = $kwset_occur_mapping[kwset_to_string($support_sets[0])];
$suppset_occur2 = $kwset_occur_mapping[kwset_to_string($support_sets[1])];
$union_length = $suppset_occur1 + $suppset_occur2 - $intersection_length;
$jaccard_index = floatval($intersection_length) / $union_length;
if ($verbose) {
echo 'processing [' . $kwset_string . '] index = ' . $jaccard_index . '<br />';
}
//check if index is no less than THRESHOLD
if ($jaccard_index >= THRESHOLD) {
//TODO: add keyword_set, uri to databse
$candidates[$kwset_string] = $keyword_set;
if ($round_count > 1) {
echo '<tr><td>' . $kwset_string . '</td><td>' . $intersection_length . '</td><td>' . $jaccard_index . '</td><td>' . kwset_to_string($support_sets[0]) . '[' . $suppset_occur1 . ']<br />' . kwset_to_string($support_sets[1]) . '[' . $suppset_occur2 . ']</td></tr>';
}
}
}
//string keys to numeric
$candidates = array_values($candidates);
//aggregate to size N+1
$current_generation = keyword_aggregate($candidates);
$round_count++;
}
echo '</table>';
}
}
mysql_close($con);
