Recommendation algorithm

Collaborative filtering recommendation system

• Ranking algorithm using likes / dislikes or star-based rating
• This package can be used in any PHP application or with any framework.
• Download package: composer require tigo/recommendation
• MIT license. Feel free to use this project. Leave a star :star: or make a fork !

If you found this project useful, consider making a donation to support the developer.

|

Getting started

Starting with composer

1. Install composer
2. Download package: composer require tigo/recommendation
3. PHP >= 7.0; Versions that have been tested: 7.2.25, 7.3.23 e 8.0.1.

//Somewhere in your project, you may need to use autoload
include __DIR__ ."/vendor/autoload.php";

Algorithms

• ranking
• euclidean
• slope one

Introduction

Recommend a product using collaborative filtering

   /**  
     $table gets the array from the database.
     $user is the foreign key that represents the user who will receive the recommendation.
   **/
   use Tigo\Recommendation\Recommend; // import class
   $client = new Recommend();
   $client->ranking($table,$user); //optional third parameter refers to the score not accepted
   $client->euclidean($table,$user); //optional third parameter refers to the minimum accepted score
   $client->slopeOne($table, $user); //optional third parameter refers to the minimum accepted score

Configuration

Sometimes, it may be necessary to rename the value of the constants (According to your database table).

• Configure: standard key (Directory: ./src/configuration/StandardKey.php)

    const SCORE = 'score'; //score  
    const PRODUCT_ID = 'product_id'; //Foreign key
    const USER_ID = 'user_id'; //Foreign key 

Example

A simple didactic demonstration of the algorithm

  /**
     Example using "rating: liked and disliked"
     like: score = 1;  dislike: score = 0
  **/
   $table = [
        ['product_id'=> 'A',
         'score'=> 1, 
         'user_id'=> 'Pedro'
        ],
        ['product_id'=> 'B',
         'score'=> 1, 
         'user_id'=> 'Pedro'
        ],
        ['product_id'=> 'A',
         'score'=> 1, 
         'user_id'=> 'João'
        ],
        ['product_id'=> 'B',
         'score'=> 1, 
         'user_id'=> 'João'
        ],
        ['product_id'=> 'C',
         'score'=> 1, 
         'user_id'=> 'João'
        ]
  ];
  use Tigo\Recommendation\Recommend; // import class
  $client = new Recommend();
  print_r($client->ranking($table,"Pedro")); // result = ['C' => 2] 
  print_r($client->ranking($table,"Pedro",1)); // result = []; 
  
  print_r($client->euclidean($table,"Pedro")); // result = ['C' => 1]
  print_r($client->euclidean($table,"Pedro", 2)); // result = [] ;  
  
  print_r($client->slopeOne($table,'Pedro')); // result = ['C' => 1]
  print_r($client->slopeOne($table,'Pedro', 2)); // result = []

Supporting this project

If you are interested in supporting this project, you can help in many ways. Leave a star :star: or make a donation of any value.

Sponsor supporting this project

• []

Contributors

• Tiago A C Pereira
• list of contributors

License

MIT license. See the archive License