Context Navigation

bayesfactory.class.php @ 5410

Revision 5410, 12.1 KB checked in by gcroes, 13 months ago (diff)
Exception => CopixException
Property svn:eol-style set to `native`

Line
1	<?php
2	/**
3	* @package tools
4	* @subpackage bayes
5	* @author Patrice Ferlet - <metal3d@copix.org>
6	* @copyright CopixTeam
7	* @licence GNU/GPL
8	*/
9
10	interface IBayesMethods {
11	public function train($pCat,$pText);
12	public function untrain($pCat,$pText);
13	public function getProba($A,$B);
14	public function setCategoriesProbas();
15	}
16
17	/**
18	* CopixBayes Class
19	* Used to get Bayesian values of probabilities
20	*
21	* @package tools
22	* @subpackage bayes
23	*/
24	class BayesFactory {
25
26	/**
27	* Get Bayes object from datamode
28	*
29	* @param const $datamode
30	* @param string $dataset name
31	* @param string $connectionname
32	* @return Bayes object cast for datamode
33	*/
34	public function create($datamode = Bayes::STATIC_DATAMODE,$dataset=null,$connectionname=null){
35	if($datamode == Bayes::DB_DATAMODE){
36	$b = new DBBayes();
37	}
38	else{
39	$b = new StaticBayes();
40	}
41	$b->setDataMode($datamode,$dataset,$connectionname);
42	return $b;
43	}
44
45	}
46
47
48	abstract class Bayes implements IBayesMethods {
49	/**
50	* Array of categories
51	*
52	* @var array
53	*/
54	protected $categories = array();
55
56	/**
57	* Weight for all categories, in fact it's the whole data count
58	*
59	* @var integer
60	*/
61	protected $numcat=0;
62
63	/**
64	* Bayes mode, database or static
65	*
66	* @var const int
67	*/
68	protected $mode=self::STATIC_DATAMODE;
69
70	/**
71	* Dataset name for database mode
72	*
73	* @var string
74	*/
75	protected $dataset="";
76
77	/**
78	* Static mode
79	*
80	*/
81	const STATIC_DATAMODE = 0;
82
83	/**
84	* Database mode
85	*
86	*/
87	const DB_DATAMODE = 1;
88
89	/**
90	* Connection name to strore dataset
91	*
92	* @var string
93	*/
94	protected $connectionName = null;
95
96	/**
97	* True if calcul is done
98	*
99	* @var boolean
100	*/
101	private $calculated = false;
102
103	/**
104	* Human readable operation done
105	*
106	* @var string
107	*/
108	private $operation = "";
109
110	/**
111	* Constructor
112	*
113	*/
114	public function __construct(){
115	$this->setDataMode();
116	}
117
118
119	/**
120	* Set the data mode
121	* @param string $dataset_name
122	* @param string $mode Bayes::STATIC_DATAMODE or Bayes::DB_DATAMODE
123	* @param string connection_name
124	*/
125	public function setDataMode($pMode=self::STATIC_DATAMODE ,$pDataset=null,$pConnectionName=null){
126	if(empty($pDataset) && $pMode==self::DB_DATAMODE){
127	throw (new CopixException("No dataset given for CopixBayesian datas used with database"));
128	}
129	$this->dataset="bayesiantable_".$pDataset;
130	$this->mode = ($pMode==self::DB_DATAMODE) ? self::DB_DATAMODE :self::STATIC_DATAMODE;
131
132	//get tables
133	if($this->mode==self::DB_DATAMODE){
134	$this->connectionName = $pConnectionName;
135	$ct = CopixDB::getConnection($pConnectionName);
136	if(!in_array($this->dataset,$ct->getTableList())){
137	if($ct instanceof CopixDBConnectionMySQL \|\| $ct instanceof CopixDBConnectionPDO_MySQL){
138	$sql = CopixFile::read(dirname(__FILE__)."/../install/template_scripts/install.pdo_mysql.sql");
139	}
140	elseif($ct instanceof CopixDBConnectionPDO_SQLite){
141	$sql = CopixFile::read(dirname(__FILE__)."/../install/template_scripts/install.pdo_sqlite.sql");
142	}
143	elseif($ct instanceof CopixDBConnectionPDO_PgSQL){
144	$sql = CopixFile::read(dirname(__FILE__)."/../install/template_scripts/install.pdo_pgsql.sql");
145	}
146	else{
147	//throw new CopixException("Data type: ".get_class($ct)." not currently supported");
148	}
149	$sql = str_replace('%TABLENAME%',$this->dataset,$sql);
150	_doQuery($sql,array(),$pConnectionName);
151	}
152	}
153
154	}
155
156
157
158	/**
159	* Get the Bayesian value for a category
160	*
161	* @param string $category
162	* @param string $data_to_test
163	* @param boolean $naive to give same chance for each category
164	* @return float $bayesian_value (in percent %)
165	*/
166	public function getBayes($B,$A,$pNaive=false){
167	$A = strtolower($A);
168	$this->setCategoriesProbas();
169	//P(B\|A)
170	//numerator: P(B) * P(A\|B)...
171	//P(B) => getCategoriesProbas for B
172	//P(A\|B) => find A in B
173	if(!isset($this->categories[$B])){
174	return 0;
175	}
176	if(!$pNaive){
177	$PB = $this->categories[$B]->percent;
178	} else {
179	$PB = (100/count($this->categories));
180	}
181	$PAB = $this->getProba($A,$B);
182	$numerator = $PB * $PAB;
183	$this->operation = $PB.'*'.$PAB;
184	$this->operation .="\n";
185	//denominator: ( P(B)(P(A\|B) + P(B2)P(A\|B2) + ... )
186	//so for every categories, we have to check
187	$den = 0;
188	foreach($this->categories as $name=>$B){
189	if(!$pNaive){
190	$PB = $B->percent;
191	}
192	$PAB = $this->getProba($A,$name);
193	$den += ($PB*$PAB);
194	$this->operation.= '+('.$PB.'*'.$PAB.')';
195	}
196
197	if($den==0) return 0;
198	$this->calculated = true;
199	return 100*$numerator/$den;
200	}
201
202
203	public function getOperation(){
204	if(!$this->calculated){
205	throw new CopixException("Calcul is not done yet");
206	}
207	$op = $this->operation;
208	list($num,$den) = explode("\n",$this->operation);
209	$den = preg_replace('/^\+/','',$den);
210	return '<table>
211	<tr>
212	<td style="text-align: center; border-bottom: 1px solid #000">'.$num.'</td>
213	</tr>
214	<tr><td "text-align: center;">'.$den.'
215	</tr></td>
216	</table>';
217	}
218
219
220	/**
221	* Prepare and split text to be registered into dataset
222	*
223	* @param strign $pText
224	* @return array of strings
225	*/
226	protected function prepareText ($pText){
227	$text = $this->remove_accents($pText);
228	$t=preg_split('/\W/is',$text);
229	$texts=array();
230	foreach ($t as $text){
231	if(strlen(trim($text))>0){
232	$texts[]=strtolower(trim($text));
233	}
234	}
235	return $texts;
236	}
237
238	/**
239	* By derernst at gmx dot ch: http://fr3.php.net/manual/fr/function.strtr.php#56973
240	*/
241	protected function remove_accents($pString, $german=false) {
242	// Single letters
243	//$string=utf8_encode($string);
244	$single_fr = explode(" ", "Ã Ã Ã Ã Ã Ã Ą Ă Ã Ć Č Ď Đ Ã Ã Ã Ã Ã Ę Ě Ğ Ã Ã Ã Ã İ Ł Ľ Ĺ Ã Ń Ň Ã Ã Ã Ã Ã Ã Ő Ŕ Ř Å Ś Ş Ť Ţ Ã Ã Ã Ã Ů Ű Ã Åœ Ź Ż Ã Ã¡ Ã¢ Ã£ Ã€ Ã¥ ą ă Ã§ ć č ď đ Ãš Ã© Ãª Ã« ę ě ğ Ã¬ Ã Ã® Ã¯ ı ł ľ ĺ Ã± ń ň Ã° Ã² Ã³ ÃŽ Ãµ Ã¶ Ãž ő ŕ ř ś Å¡ ş ť ţ Ã¹ Ãº Ã» ÃŒ ů ű Ãœ Ã¿ ÅŸ ź ż");
245	$single_to = explode(" ", "A A A A A A A A C C C D D D E E E E E E G I I I I I L L L N N N O O O O O O O R R S S S T T U U U U U U Y Z Z Z a a a a a a a a c c c d d e e e e e e g i i i i i l l l n n n o o o o o o o o r r s s s t t u u u u u u y y z z z");
246	$single = array();
247	for ($i=0; $i<count($single_fr); $i++) {
248	$single[$single_fr[$i]] = $single_to[$i];
249	}
250	// Ligatures
251	$ligatures = array("Ã"=>"Ae", "ÃŠ"=>"ae", "Å"=>"Oe", "Å"=>"oe", "Ã"=>"ss");
252	// German umlauts
253	$umlauts = array("Ã"=>"Ae", "Ã€"=>"ae", "Ã"=>"Oe", "Ã¶"=>"oe", "Ã"=>"Ue", "ÃŒ"=>"ue");
254	// Replace
255	$replacements = array_merge($single, $ligatures);
256	if ($german) $replacements = array_merge($replacements, $umlauts);
257	$pString = strtr($pString, $replacements);
258	return $pString;
259	}
260
261
262
263	/**
264	* Add category to the data set
265	*
266	* @param unknown_type $cat
267	*/
268	protected function addCategory ($pCat){
269	$this->categories[$pCat]=new stdClass;
270	$this->categories[$pCat]->texts=array();
271	$this->categories[$pCat]->counter=0;
272	$this->categories[$pCat]->percent = 0;
273	}
274	}
275
276	/**
277	* StaticBayes is a "one shot" class to process datas and
278	* get bayesian probabilities
279	*
280	*/
281	class StaticBayes extends Bayes {
282
283	/**
284	* Prepare categories
285	*
286	*/
287	public function setCategoriesProbas(){
288	foreach($this->categories as $name=>$cat){
289	$cat->percent = $cat->counter * 100 / $this->numcat;
290	}
291
292	}
293
294	/**
295	* Get probability of A is in B
296	*
297	* @param string $category category name
298	* @param string $data_to_check data to check
299	* @return float $proba (in percent %)
300	*/
301	public function getProba($A,$B){
302	$A = $this->remove_accents($A);
303	$A = preg_split('/\W/is',$A);
304	$numwords = 0;
305	$found = 0;
306	foreach($this->categories[$B]->texts as $words){
307	$numwords += count($words);
308	foreach($words as $word){
309	foreach($A as $find){
310	if($find==$word){
311	$found++;
312	}
313	}
314	}
315	}
316
317	return $found * 100 / $numwords;
318	}
319
320	/**
321	* Add/update category and datas for this category
322	*
323	*
324	* @param string $category
325	* @param string $text
326	*/
327	public function train($pCat,$pText){
328	$texts = $this->prepareText($pText);
329
330	if(!isset($this->categories[$pCat])) $this->addCategory($pCat);
331
332	$this->categories[$pCat]->texts[]=$texts;
333	$this->categories[$pCat]->counter+=count($texts);
334	$this->numcat+=count($texts);
335
336	}
337
338	/**
339	* Untrain/remove the data from dataset
340	*
341	* @param string $categoryname
342	* @param string $text
343	*/
344
345	public function untrain($pCat,$pText){
346	$texts = $this->prepareText($pText);
347
348	if(isset($this->categories[$pCat]) && isset($this->categories[$pCat]->texts)){
349	$i=0;
350	foreach($this->categories[$pCat]->texts as $t){
351	if($texts == $t){
352	//remove one word for numcat:
353	$this->numcat--;
354	$this->categories[$cat]->counter--;
355	unset($this->categories[$pCat]->texts[$i]);
356	break;
357	}
358	$i++;
359	}
360	if(count($this->categories[$pCat]->texts)<1){
361	unset($this->categories[$pCat]);
362	}
363	}
364
365	}
366
367	}
368
369
370	/**
371	* DBBayes store datas into database
372	*
373	*/
374	class DBBayes extends Bayes {
375
376	/**
377	* Prepare categories
378	*
379	*/
380	public function setCategoriesProbas(){
381	//categories counter
382	$cats = _doQuery('select distinct category_bayes from '.$this->dataset,array(),$this->connectionName);
383
384	//words counter
385	$acount = _doQuery("select sum(numdatas_bayes) count from ".$this->dataset,array(),$this->connectionName);
386
387	foreach($cats as $cat){
388	$this->addCategory($cat->category_bayes);
389	$count = _doQuery("select sum(numdatas_bayes) count from ".$this->dataset." where category_bayes='".$cat->category_bayes."'",array(),$this->connectionName);
390	$this->categories[$cat->category_bayes]->percent= $count[0]->count * 100 / $acount[0]->count;
391
392	$name=$cat->category_bayes;
393	}
394	}
395
396	/**
397	* Get probability of A is in B
398	*
399	*
400	* @param string $category category name
401	* @param string $data_to_check data to check
402	* @return float $proba (in percent %)
403	*/
404	public function getProba($A,$B){
405	$A = $this->remove_accents($A);
406	$A = preg_split('/\W/is',$A);
407	$numwords = 0;
408	$found = 0;
409
410	$numwords = _doQuery('select sum (numdatas_bayes) numwords from '.$this->dataset.' where category_bayes="'.$B.'"',array(),$this->connectionName);
411	$numwords = $numwords[0]->numwords;
412	foreach($A as $find){
413	$sets = _doQuery("select numdatas_bayes from ".$this->dataset." where category_bayes=\"$B\" AND datas_bayes ='$find'",array(),$this->connectionName);
414	foreach($sets as $set){
415	$found += $set->numdatas_bayes;
416	}
417	}
418	return $found * 100 / $numwords;
419	}
420
421	/**
422	* Add/update category and datas for this category
423	*
424	*
425	* @param string $category
426	* @param string $text
427	*/
428	public function train($pCat,$pText){
429	$texts = $this->prepareText($pText);
430	foreach ($texts as $word){
431	$res = _ioDao($this->dataset,$this->connectionName)->findBy(_daoSP()
432	->addCondition('datas_bayes','=',$word)
433	->addCondition('category_bayes','=',$pCat)
434	->addCondition('dataset_bayes','=',$this->dataset)
435	,$this->connectionName);
436	$method="update";
437	if (count($res)<1 \|\| !isset($res[0])) {
438	$rec = CopixDAOFactory::createRecord($this->dataset,$this->connectionName,$this->connectionName);
439	$rec->category_bayes = $pCat;
440	$rec->numdatas_bayes = 1;
441	$rec->datas_bayes = $word;
442	$rec->dataset_bayes=$this->dataset;
443	$method = "insert";
444	} else {
445	$rec = $res[0];
446	$rec->numdatas_bayes ++;
447	}
448	_ioDao($this->dataset,$this->connectionName)->$method($rec);
449	}
450
451	}
452
453	/**
454	* Untrain, remove the data from dataset
455	*
456	* @param string $categoryname
457	* @param string $text
458	*/
459	public function untrain($pCat,$pText){
460	$texts = $this->prepareText($pText);
461
462	foreach($texts as $text){
463	$rec = _ioDao($this->dataset,$this->connectionName)->findBy(_daoSp()
464	->addCondition('datas_bayes','=',$text)
465	->addCondition('category_bayes','=',$pCat)
466	);
467
468	if(count($rec)){
469	//decrements the data number for this word
470	if($rec[0]->numdatas_bayes>1) {
471	$rec[0]->numdatas_bayes--;
472	_ioDao($this->dataset,$this->connectionName)->update($rec[0]);
473	}else{
474	//this word have to be forget
475	_ioDao($this->dataset,$this->connectionName)->delete($rec[0]->id_bayes);
476	}
477	}
478	}
479
480
481	}
482
483
484	}

Note: See TracBrowser for help on using the repository browser.

Context Navigation

source: trunk/project/modules/stable/tools/bayes/classes/bayesfactory.class.php @ 5410

Download in other formats: