/**
* @private
*/
function getSpamProbability($blogId, $topic, $text, $userName, $userEmail, $userUrl)
{
$tokenizer = new BayesianTokenizer();
$tokensTopic = $tokenizer->addContextMark($tokenizer->tokenize($topic), TOKEN_TOPIC_MARK);
$tokensText = $tokenizer->tokenize($text);
$tokensUserName = $tokenizer->addContextMark($tokenizer->tokenize($userName), TOKEN_USER_NAME_MARK);
$tokensUserEmail = $tokenizer->addContextMark($tokenizer->tokenize($userEmail), TOKEN_USER_EMAIL_MARK);
$tokensUserUrl = $tokenizer->addContextMark($tokenizer->tokenize($userUrl), TOKEN_USER_URL_MARK);
$tokens = array_merge($tokensTopic, $tokensText, $tokensUserName, $tokensUserEmail, $tokensUserUrl);
$significantTokens = BayesianFilter::_getMostSignificantTokens($blogId, $tokens);
return BayesianFilter::_getBayesProbability($significantTokens);
}
/**
* untrains the filter
*
* @param blogId The blog id
* @param topic The topic of the comment/article that we're using to untrain the filter
* @param text The text of the comment/articles that we're usingn to untrain the filter
* @param userName Name of the user posting this comment/article
* @param userEmail Email address of the user posting this comment/article
* @param userUrl URL of the user posting this comment/article
* @param spam Wether we should unmark these contents as spam or not. The content will be unmarked
* as non-spam by default
* @static
* @see train
*/
function untrain($blogId, $topic, $text, $userName, $userEmail, $userUrl, $spam = false)
{
$tokenizer = new BayesianTokenizer();
$tokensTopic = $tokenizer->addContextMark($tokenizer->tokenize($topic), TOKEN_TOPIC_MARK);
$tokensText = $tokenizer->tokenize($text);
$tokensUserName = $tokenizer->addContextMark($tokenizer->tokenize($userName), TOKEN_USER_NAME_MARK);
$tokensUserEmail = $tokenizer->addContextMark($tokenizer->tokenize($userEmail), TOKEN_USER_EMAIL_MARK);
$tokensUserUrl = $tokenizer->addContextMark($tokenizer->tokenize($userUrl), TOKEN_USER_URL_MARK);
$tokens = array_merge($tokensTopic, $tokensText, $tokensUserName, $tokensUserEmail, $tokensUserUrl);
$bayesianFilterInfos = new BayesianFilterInfos();
$bayesianFilterInfo = $bayesianFilterInfos->getBlogBayesianFilterInfo($blogId);
$totalSpam = $bayesianFilterInfo->getTotalSpam();
$totalNonSpam = $bayesianFilterInfo->getTotalNonSpam();
$bayesianTokens = new BayesianTokens();
if ($spam) {
$bayesianTokens->decSpamOccurrencesFromTokensArray($blogId, $tokens, $totalSpam, $totalNonSpam);
$bayesianFilterInfos->decTotalSpam($bayesianFilterInfo->getId());
} else {
$bayesianTokens->decNonSpamOccurrencesFromTokensArray($blogId, $tokens, $totalSpam, $totalNonSpam);
$bayesianFilterInfos->decTotalNonSpam($bayesianFilterInfo->getId());
}
return true;
}
