echo '<table border="1px"><tr><th>Keyword Set</th><th>Occurrence</th><th>Jaccard Index</th><th>Support sets</th></tr>';
while (count($current_generation) > 0) {
$candidates = array();
foreach ($current_generation as $group) {
$keyword_set = $group[0];
$kwset_string = kwset_to_string($keyword_set);
//avoid unnecessary typing
$support_sets = $group[1];
//if keyword set already in the support set
//skip it
if (array_key_exists($kwset_string, $candidates)) {
continue;
}
//in case the occurrence is not calculated
if (!array_key_exists($kwset_string, $kwset_occur_mapping)) {
$occur = occurrence($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) {
$candidates[$kwset_string] = $keyword_set;
if ($round_count > 1) {
}
//echo '</table>';
echo "entries with keywords/total entries: {$count} / {$TOTAL}";
$all = array_unique($all);
//sort($all);
$keyword_occr = array();
foreach ($all as $keyword) {
$occurrence = occurrence(array($keyword), $all_keywords_arr);
$keyword_occr[$keyword] = $occurrence;
}
arsort($keyword_occr);
echo '<table border="1px"><tr><th>Keywords of size 2</th><th>occurrence</th><th>ratio(=occur/max(sub_occur))</th></tr>';
foreach ($all_keywords_arr as $keywords_arr) {
$subsets_2 = generate_next($keywords_arr, expand_dimension($keywords_arr));
foreach ($subsets_2 as $subset_2) {
$occur_2 = occurrence($subset_2, $all_keywords_arr);
//print elements in subset_2
echo '<tr><td>';
$max_elem_occr = 0;
foreach ($subset_2 as $elem) {
echo $elem . "({$keyword_occr[$elem]}) ";
$max_elem_occr = max($max_elem_occr, $keyword_occr[$elem]);
}
$ratio = floatval($occur_2) / $max_elem_occr;
//print occurrence and ratio
echo "</td><td>{$occur_2}</td><td>{$ratio}</td></tr>";
}
//split sets
echo '<tr><td> </td><td> </td></tr>';
}
echo '</table>';