Case 2: Estimating object relevance

Implementation in pure PHP

In this variant we estimate object relevance via the dot product of features and weights. It is a fast and transparent baseline for ranking. We define a feature vector and an importance vector, then compute the final score as the sum of pairwise products.

Example of code


<?php

function dotProduct(array $a, array $b): float {
    $n = count($a);

    if ($n !== count($b)) {
        throw new InvalidArgumentException('Vectors must have the same length');
    }

    $sum = 0.0;

    for ($i = 0; $i < $n; $i++) {
        $sum += $a[$i] * $b[$i];
    }

    return $sum;
}

Implementation in RubixML

In the second variant we use RubixML with a Ridge regressor trained on examples with known target values. After training, the model predicts relevance for a new object, which is useful when moving from manual weights to a trainable approach.

Example of code


<?php

use Rubix\ML\Datasets\Labeled;
use Rubix\ML\Regressors\Ridge;

$samples = [
    [10, 5, 2],
    [4, 1, 0],
    [20, 8, 5],
];

$labels = [8, 2, 15];

$dataset = new Labeled($samples, $labels);

$model = new Ridge(1.0);
$model->train($dataset);