| 1 |
<?php
|
| 2 |
// $Id$
|
| 3 |
|
| 4 |
/**
|
| 5 |
* @file Relevance feedback module
|
| 6 |
*/
|
| 7 |
|
| 8 |
/**
|
| 9 |
* Implementation of hook_init()
|
| 10 |
*/
|
| 11 |
function relevance_feedback_init() {
|
| 12 |
if (arg(0) == 'search' && arg(1) == 'node') {
|
| 13 |
if (variable_get('relevance_feedback_type', 'explicit') == 'implicit') {
|
| 14 |
drupal_add_js(drupal_get_path('module', 'relevance_feedback'). '/relevance_feedback.implicit.js', 'module');
|
| 15 |
}
|
| 16 |
else if (variable_get('relevance_feedback_type', 'explicit') == 'explicit') {
|
| 17 |
drupal_add_js(drupal_get_path('module', 'relevance_feedback'). '/relevance_feedback.explicit.js', 'module');
|
| 18 |
drupal_add_css(drupal_get_path('module', 'relevance_feedback'). '/relevance_feedback.css', 'module');
|
| 19 |
}
|
| 20 |
}
|
| 21 |
}
|
| 22 |
|
| 23 |
/**
|
| 24 |
* Implementation of hook_menu()
|
| 25 |
*/
|
| 26 |
function relevance_feedback_menu() {
|
| 27 |
$items = array();
|
| 28 |
$items['relevance_feedback/js'] = array(
|
| 29 |
'page callback' => 'relevance_feedback_js',
|
| 30 |
'access arguments' => array('access content'),
|
| 31 |
'type' => MENU_CALLBACK,
|
| 32 |
);
|
| 33 |
$items['admin/settings/relevance_feedback'] = array(
|
| 34 |
'title' => 'Relevance Feedback',
|
| 35 |
'description' => 'Configure relevance feedback.',
|
| 36 |
'page callback' => 'drupal_get_form',
|
| 37 |
'page arguments' => array('relevance_feedback_settings'),
|
| 38 |
'access arguments' => array('administer site configuration'),
|
| 39 |
'file' => 'relevance_feedback.admin.inc',
|
| 40 |
);
|
| 41 |
return $items;
|
| 42 |
}
|
| 43 |
|
| 44 |
function relevance_feedback_js() {
|
| 45 |
|
| 46 |
$search = relevance_feedback_get_search();
|
| 47 |
|
| 48 |
if (isset($search['keys']) && $search['keys'] == $_GET['keys']) {
|
| 49 |
if (isset($_GET['url'])) {
|
| 50 |
$url = parse_url($_GET['url']);
|
| 51 |
$path = drupal_get_normal_path($url['path']);
|
| 52 |
if (strpos($path, '/node') === 0) {
|
| 53 |
$nid = substr($path, 6);
|
| 54 |
}
|
| 55 |
}
|
| 56 |
else if (isset($_GET['nid'])) {
|
| 57 |
$nid = $_GET['nid'];
|
| 58 |
}
|
| 59 |
|
| 60 |
if (isset($nid) && is_numeric($nid) && isset($_GET['type'])) {
|
| 61 |
if ($_GET['type'] == 'rel') {
|
| 62 |
relevance_feedback_add_relevant_result($nid);
|
| 63 |
}
|
| 64 |
else if ($_GET['type'] == 'nonrel') {
|
| 65 |
relevance_feedback_add_nonrelevant_result($nid);
|
| 66 |
}
|
| 67 |
else {
|
| 68 |
relevance_feedback_clear_result_status($nid);
|
| 69 |
}
|
| 70 |
relevance_feedback_process();
|
| 71 |
print isset($_SESSION['relevance_feedback']['refresh']) ? 1 : '';
|
| 72 |
}
|
| 73 |
}
|
| 74 |
}
|
| 75 |
|
| 76 |
|
| 77 |
/**
|
| 78 |
* Implementation of hook_nodeapi()
|
| 79 |
*/
|
| 80 |
function relevance_feedback_nodeapi($node, $op) {
|
| 81 |
if ($op == 'search result' && variable_get('relevance_feedback_type', 'explicit') == 'explicit') {
|
| 82 |
$path = drupal_get_path('module', 'relevance_feedback');
|
| 83 |
$output = theme('image', $path .'/images/up.gif', 'up', t('Mark this result as relevant.'), array('class' => 'relevance up'));
|
| 84 |
$output .= theme('image', $path .'/images/down.gif', 'down', t('Mark this result as non-relevant.'), array('class' => 'relevance down'));
|
| 85 |
$status = relevance_feedback_get_result_status($node->nid);
|
| 86 |
$status = $status ? 'rf_status = "1" ' : ($status === FALSE ? 'rf_status = "0" ' : '');
|
| 87 |
return '<span class="relevance_feedback" id="relevance_feedback_'. $node->nid .'" '. $status .'>'. $output. '</span>';
|
| 88 |
}
|
| 89 |
}
|
| 90 |
|
| 91 |
/**
|
| 92 |
* Implementation of hook_form_alter()
|
| 93 |
*/
|
| 94 |
function relevance_feedback_form_alter(&$form, &$form_state, $form_id) {
|
| 95 |
if ($form_id == 'search_form' && $form['module']['#value'] == 'node') {
|
| 96 |
if ($form_state['post']) {
|
| 97 |
$form['#submit'][] = 'relevance_feedback_init_search';
|
| 98 |
//relevance_feedback_init_search($form_state['post']['keys']);
|
| 99 |
}
|
| 100 |
else {
|
| 101 |
$search = relevance_feedback_get_search();
|
| 102 |
$value = $form['basic']['inline']['keys']['#default_value'];
|
| 103 |
if ($search && $value && $search['keys']) {
|
| 104 |
//show the original search query to the user
|
| 105 |
$form['basic']['inline']['keys']['#default_value'] = $search['keys'];
|
| 106 |
//refresh if necessary
|
| 107 |
if ($query = relevance_feedback_get_new_search_query()) {
|
| 108 |
unset($_SESSION['relevance_feedback']['refresh']);
|
| 109 |
drupal_set_message(t('Your search query was refined for better search results.'));
|
| 110 |
drupal_goto('search/node/'. $query);
|
| 111 |
}
|
| 112 |
}
|
| 113 |
}
|
| 114 |
}
|
| 115 |
}
|
| 116 |
|
| 117 |
function relevance_feedback_init_search($form, $form_state) {
|
| 118 |
$keys = $form_state['values']['processed_keys'];
|
| 119 |
$_SESSION['relevance_feedback'] = array();
|
| 120 |
$_SESSION['relevance_feedback']['keys'] = $keys;
|
| 121 |
$_SESSION['relevance_feedback']['query'] = $keys;
|
| 122 |
drupal_set_message("Detected new search: ". check_plain($keys));
|
| 123 |
}
|
| 124 |
|
| 125 |
function relevance_feedback_get_search() {
|
| 126 |
return isset($_SESSION['relevance_feedback']) ? $_SESSION['relevance_feedback'] : array();
|
| 127 |
}
|
| 128 |
|
| 129 |
function relevance_feedback_add_relevant_result($nid) {
|
| 130 |
relevance_feedback_clear_result_status($nid);
|
| 131 |
$_SESSION['relevance_feedback'] += array('rel' => array());
|
| 132 |
$_SESSION['relevance_feedback']['rel'][$nid] = $nid;
|
| 133 |
}
|
| 134 |
|
| 135 |
function relevance_feedback_add_nonrelevant_result($nid) {
|
| 136 |
relevance_feedback_clear_result_status($nid);
|
| 137 |
$_SESSION['relevance_feedback'] += array('nonrel' => array());
|
| 138 |
$_SESSION['relevance_feedback']['nonrel'][$nid] = $nid;
|
| 139 |
}
|
| 140 |
|
| 141 |
function relevance_feedback_clear_result_status($nid) {
|
| 142 |
if (isset($_SESSION['relevance_feedback']['nonrel'])) {
|
| 143 |
unset($_SESSION['relevance_feedback']['nonrel'][$nid]);
|
| 144 |
}
|
| 145 |
if (isset($_SESSION['relevance_feedback']['rel'])) {
|
| 146 |
unset($_SESSION['relevance_feedback']['rel'][$nid]);
|
| 147 |
}
|
| 148 |
}
|
| 149 |
|
| 150 |
/**
|
| 151 |
* Status of a result:
|
| 152 |
* true -> relevant
|
| 153 |
* false -> non relevant
|
| 154 |
* null -> undefined
|
| 155 |
*/
|
| 156 |
function relevance_feedback_get_result_status($nid) {
|
| 157 |
if (isset($_SESSION['relevance_feedback']['rel'][$nid])) {
|
| 158 |
return TRUE;
|
| 159 |
}
|
| 160 |
else if (isset($_SESSION['relevance_feedback']['nonrel'][$nid])) {
|
| 161 |
return FALSE;
|
| 162 |
}
|
| 163 |
return NULL;
|
| 164 |
}
|
| 165 |
|
| 166 |
/**
|
| 167 |
* Use the algorithm on the gathered data
|
| 168 |
*/
|
| 169 |
function relevance_feedback_process() {
|
| 170 |
unset($_SESSION['relevance_feedback']['refresh']);
|
| 171 |
|
| 172 |
$_SESSION['relevance_feedback'] += array('rel' => array(), 'nonrel' => array(), 'used' => 0);
|
| 173 |
$count = count($_SESSION['relevance_feedback']['rel']) - $_SESSION['relevance_feedback']['used'];
|
| 174 |
|
| 175 |
if ($count >= variable_get('relevance_feedback_count', 5)) {
|
| 176 |
$_SESSION['relevance_feedback']['used'] += variable_get('relevance_feedback_count', 5);
|
| 177 |
|
| 178 |
// calculate new query here.
|
| 179 |
$query = relevance_feedback_get_refined_query();
|
| 180 |
|
| 181 |
relevance_feedback_process_set_new_search_query($query);
|
| 182 |
}
|
| 183 |
}
|
| 184 |
|
| 185 |
/**
|
| 186 |
* Sets a new search query
|
| 187 |
*/
|
| 188 |
function relevance_feedback_process_set_new_search_query($query) {
|
| 189 |
$_SESSION['relevance_feedback']['refresh'] = TRUE;
|
| 190 |
$_SESSION['relevance_feedback']['query'] = $query;
|
| 191 |
}
|
| 192 |
|
| 193 |
function relevance_feedback_get_new_search_query() {
|
| 194 |
return isset($_SESSION['relevance_feedback']['refresh']) ? $_SESSION['relevance_feedback']['query'] : FALSE;
|
| 195 |
}
|
| 196 |
|
| 197 |
/**
|
| 198 |
* Refines the query making use of the feedback
|
| 199 |
*/
|
| 200 |
function relevance_feedback_get_refined_query() {
|
| 201 |
$words = explode(' ', $_SESSION['relevance_feedback']['keys']);
|
| 202 |
|
| 203 |
// Build the original search query
|
| 204 |
$q0 = $special = array();
|
| 205 |
foreach ($words as $word) {
|
| 206 |
if (strpos($word, ':') === FALSE) {
|
| 207 |
$q0[$word] = variable_get('relevance_feedback_treshold', 0.10);
|
| 208 |
}
|
| 209 |
else {
|
| 210 |
//filter out special commands
|
| 211 |
$special[] = $word;
|
| 212 |
}
|
| 213 |
}
|
| 214 |
|
| 215 |
$function = 'relevance_feedback_apply_'. variable_get('relevance_feedback_algo', 'rocchio');
|
| 216 |
$qm = $function($q0);
|
| 217 |
|
| 218 |
|
| 219 |
relevance_feedback_debug("Original search query: <pre>".print_r($q0, 1)."</pre>");
|
| 220 |
asort($qm);
|
| 221 |
relevance_feedback_debug("Optimised query: <pre>".print_r($qm, 1)."</pre>");
|
| 222 |
|
| 223 |
// Filter out words below the treshold
|
| 224 |
$qm = array_filter($qm, '_relevance_feedback_filter');
|
| 225 |
relevance_feedback_debug("Add search keywords: <pre>".print_r($qm, 1)."</pre>");
|
| 226 |
|
| 227 |
// Return the new keywords to look for
|
| 228 |
return implode(' ', $special) . ' '. implode(' OR ', array_keys($qm));
|
| 229 |
}
|
| 230 |
|
| 231 |
function _relevance_feedback_filter($value) {
|
| 232 |
return $value >= variable_get('relevance_feedback_treshold', 0.10);
|
| 233 |
}
|
| 234 |
|
| 235 |
function relevance_feedback_debug($msg) {
|
| 236 |
if (variable_get('relevance_feedback_debug', 0)) {
|
| 237 |
drupal_set_message($msg);
|
| 238 |
}
|
| 239 |
}
|
| 240 |
|
| 241 |
function relevance_feedback_apply_rocchio($q0) {
|
| 242 |
$count_rel = count($_SESSION['relevance_feedback']['rel']);
|
| 243 |
$count_nonrel = count($_SESSION['relevance_feedback']['nonrel']);
|
| 244 |
$alpha = variable_get('relevance_feedback_rocchio_alpha', 1);
|
| 245 |
$beta = variable_get('relevance_feedback_rocchio_beta', 0.75);
|
| 246 |
$gamma = variable_get('relevance_feedback_rocchio_gamma', 0.15);
|
| 247 |
|
| 248 |
$qm = array();
|
| 249 |
|
| 250 |
// Process q0
|
| 251 |
foreach ($q0 as $word => $weight) {
|
| 252 |
$qm[$word] = $weight * $alpha;
|
| 253 |
}
|
| 254 |
|
| 255 |
// Process relevant documents
|
| 256 |
foreach ($_SESSION['relevance_feedback']['rel'] as $nid) {
|
| 257 |
// get drupal's tf idf like value for each word of the document
|
| 258 |
$result = db_query("SELECT si.word, st.count * si.score AS weight FROM {search_index} si ".
|
| 259 |
"INNER JOIN {search_total} st ON st.word = si.word WHERE si.sid = %d and si.type ='node' ", $nid);
|
| 260 |
while ($row = db_fetch_object($result)) {
|
| 261 |
$qm += array($row->word => 0);
|
| 262 |
$qm[$row->word] += $row->weight * $beta / $count_rel;
|
| 263 |
}
|
| 264 |
}
|
| 265 |
// Process non-relevant documents
|
| 266 |
foreach ($_SESSION['relevance_feedback']['nonrel'] as $nid) {
|
| 267 |
// get drupal's tf idf like value for each word of the document
|
| 268 |
$result = db_query("SELECT si.word, st.count * si.score AS weight FROM {search_index} si ".
|
| 269 |
"INNER JOIN {search_total} st ON st.word = si.word WHERE si.sid = %d and si.type ='node' ", $nid);
|
| 270 |
while ($row = db_fetch_object($result)) {
|
| 271 |
$qm += array($row->word => 0);
|
| 272 |
$qm[$row->word] -= $row->weight * $gamma / $count_nonrel;
|
| 273 |
}
|
| 274 |
}
|
| 275 |
return $qm;
|
| 276 |
}
|
| 277 |
|