README.md

OpenCF

PHP implementation of the (Weighted Slopeone,Cosine, Weighted Cosine) rating-based collaborative filtering schemes.

To learn all about it, head over to the extensive documentation.

Installation

OpenCF Package requires PHP 7.4 or higher.

INFO: If you are using an older version of php this package will not function correctly.

The supported way of installing OpenCF package is via Composer.

composer require phpjuice/opencf

Usage

OpenCF Package is designed to be very simple and straightforward to use. All you have to do is:

1. Load a training set (dataset)
2. Predict future ratings using a recommender. (Weighted Slopeone,Cosine, Weighted Cosine)

Create Recommender Service

The OpenCF recommender service is created by direct instantiation:

use OpenCF\RecommenderService;

// Create an instance
$recommenderService = new RecommenderService($dataset);

Adding dataset

Adding a dataset to the recommender can be done using the constructor or can be easily done by providing an array of users ratings via the setDataset() method:

$dataset = [
    "squid" => [
        "user1" => 1,
        "user2" => 1,
        "user3" => 0.2,
    ],
    "cuttlefish" => [
        "user1" => 0.5,
        "user3" => 0.4,
        "user4" => 0.9,
    ],
    "octopus" => [
        "user1" => 0.2,
        "user2" => 0.5,
        "user3" => 1,
        "user4" => 0.4,
    ],
    "nautilus" => [
        "user2" => 0.2,
        "user3" => 0.4,
        "user4" => 0.5,
    ],
];

$recommenderService->setDataset($dataset);

Getting Predictions

All you have to do to predict ratings for a new user is to retrieve an engine from the recommender service and & run the predict() method.

// Get a recommender
$recommender = $recommenderService->cosine(); // Cosine recommender
// OR
$recommender = $recommenderService->weightedCosine(); // WeightedCosine recommender
// OR
$recommender = $recommenderService->weightedSlopeone(); // WeightedSlopeone recommender

// Predict future ratings
$results = $recommender->predict([
    "squid" => 0.4
]);

This should produce the following results when using WeightedSlopeone recommender

[
  "cuttlefish" => 0.25,
  "octopus" => 0.23,
  "nautilus" => 0.1
];

Running the tests

you can easily run tests using composer

composer test

Built With

• PHP - The programing language used
• Composer - Dependency Management
• Pest - An elegant PHP Testing Framework

Changelog

Please see the changelog for more information on what has changed recently.

Contributing

Please see CONTRIBUTING.md for details and a todo list.

Security

If you discover any security related issues, please email author instead of using the issue tracker.

Credits

• Daniel Lemire
• SlopeOne Predictors for Online Rating-Based Collaborative Filtering
• Distance Weighted Cosine Similarity

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

License

license. Please see the Licence for more information.