PHP與機器學習：如何進行情緒分析與評論建模

use PhpmlTokenizationWhitespaceTokenizer;
use PhpmlFeatureExtractionTfIdfTransformer;
use PhpmlFeatureExtractionTokenCountVectorizer;
use PhpmlClassificationSVC;
use PhpmlSupportVectorMachineKernel;

$text = "这部电影真是太棒了！演员表现出色，剧情扣人心悬，非常推荐！";

$vectorizer = new TokenCountVectorizer(new WhitespaceTokenizer());
$tfIdfTransformer = new TfIdfTransformer();

$vectorizer->fit([$text]);
$vectorizer->transform([$text]);

$classifier = new SVC(Kernel::RBF, $cost = 1000);
$classifier->train($samples = [$text], $labels = ['positive']);

$result = $classifier->predict($vectorizer->transform([$text]));

echo $result; // 输出：positive

use PhpmlTokenizationWhitespaceTokenizer;
use PhpmlFeatureExtractionTfIdfTransformer;
use PhpmlFeatureExtractionTokenCountVectorizer;
use PhpmlClassificationNaiveBayes;

$comments = [
    '这家餐厅的食物非常好吃，服务也很好！',
    '这个产品真的很好，质量非常出色！',
    '这本书真是一本好书，非常推荐阅读！',
    '这个电影太糟糕了，不值得一看！'
];

$labels = ['positive', 'positive', 'positive', 'negative'];

$vectorizer = new TokenCountVectorizer(new WhitespaceTokenizer());
$tfIdfTransformer = new TfIdfTransformer();

$vectorizer->fit($comments);
$vectorizer->transform($comments);

$classifier = new NaiveBayes();
$classifier->train($vectorizer->transform($comments), $labels);

$newComment = '这个产品质量太差，根本不能用！';

$result = $classifier->predict($vectorizer->transform([$newComment]));

echo $result; // 输出：negative