function countWordFreq($dataset_file, $wordbank_file = null)
{
if (!$wordbank_file) {
$wordbank_file = "w." . $dataset_file;
}
readWordBankFromJSON($wordbank_file);
$fp = fopen($dataset_file . ".csv", "r");
while (!feof($fp)) {
$tweet = fgetcsv($fp, 256);
if ($tweet) {
$tweet_words = featureExtractor($tweet[1]);
if ($tweet_words) {
updatewordBank($tweet_words, $tweet[0]);
}
}
}
writeWordBankToJSON($wordbank_file);
}
function classify($tweet)
{
//readWordBankFromJSON();
$tweet_words = featureExtractor($tweet);
$tweetWord_freq = getTweetWordFreq($tweet_words);
$result["positive"] = 1;
$result["negative"] = 1;
foreach ($tweetWord_freq as $word => $word_freq) {
//if($word_freq["positive"]!=0)
$result["positive"] *= $word_freq["positive"] / $word_freq["total"];
//if($word_freq["negative"]!=0)
$result["negative"] *= $word_freq["negative"] / $word_freq["total"];
//$result["positive"]*=10;
//$result["negative"]*=10;
//print($result["positive"]." : ".$result["negative"].", ");
}
//print("\n");
return $result;
}