Hybrid recommendation system combined content-based filtering and collaborative prediction using artificial neural network

Abstract

Recommendation systems are information filtering tools that present items to users based on their preferences and behavior, for example, suggestions about scientific papers or music a user might like. Based on what we said and with the development of computer science that has started to take an interest in big data and how it is used to discover user interest, we have found a lot of research going on in the area of recommendation and there are powerful systems available. In the unsupervised learning domain, this paper introduces a novel method for creating a hybrid recommender framework that combines Collaborative Filtering with Content Based Approach and Self-Organizing Map neural network technique. By testing our system on a subset of the Movies Database, we demonstrate that our method outperforms state-of-the-art methods in terms of accuracy and precision, as well as improving the efficiency of the traditional Collaborative Filtering methodology.

Introduction

Nowadays, with the ever increasing volume, complexity and availability of online information, recommendation systems have been an effective solution to overcome such information overload [1][2].

The main objective of a recommender system is to predict user preferences, in another way system that offer specific items to users among a large number of choices according to their tastes. Suggestions for movies on Netflix, or products on Amazon, or Videos on YouTube, are real-world examples of using recommender systems in our life [3].

In the mid-1990s, recommendation systems became a famous topic of research. Awareness in recommendation systems has grown significantly in recent years, and recommendation systems now play a significant role in commercial websites and well-known businesses such as Spotify, Facebook, LinkedIn, and IMdb $\dots$ because these systems help them to increase the number of items sold and sell more diversified items or even increase the user satisfaction. This demand for this type of systems has given the green light to researchers to develop powerful systems and several researches have been carried out in this field [4].

A recommendation method must demonstrate that a product needs to be recommended in order to find relevant products for the user. There are many kinds of recommendation algorithm approaches for this, the most prominent of which are collaborative filtering, content-based filtering, and hybrid systems.

Recommendations to the active client in the Collaborative Filtering approach are focused on items/products that other users with common preferences have enjoyed in the past. Many users’ taste similarity is determined based on the similarity of ranking data (scale of 1 to 5 for movies) or the users’ browsing history. The Content-Based model suggests products to the active user based on items he has already enjoyed. The resemblance of the products is determined by the characteristics associated with the compared items, such as the context, title, or even the product picture. For example, if a user gives a song in the HipHop genre a favorable rating, the machine can learn to recommend other songs in that genre. Hybrid methods [5][6] are built on combining existing strategies to benefit from the advantages of both and the fewest disadvantages. The most well-known and widely used solution is to incorporate Collaborative Filtering with other approaches [7].

We previously suggested a novel intelligent recommendation system [8] that integrates collaborative filtering and K-means clustering in our earlier work. Also, when items (movies) are grouped by genre attributes using K-means and users are categorized based on their item preferences and the genres they prefer to watch, we employed specific user demographic attributes such as gender and age to construct segmented user profiles [9]. Then The Collaborative Filtering technique is used to the cluster where the user belongs to recommend items to an active user.

In this paper, we propose a powerful recommendation method based on a Hybrid approach that blends Collaborative Filtering with Content-Based and is supervised by the ranking list provided by the self-organizing map, a well-known unsupervised learning artificial neural network technique.

The following is a summary of the paper’s structure. The following section discusses related work in the recommendation area. The adaptive solution and the various steps to construct our system are presented in Section 2. Sections 3 Proposed work, 4 Result and discussion describe the methods used to compare our system to other state-of-the-art approaches in the field of recommendation and analyze the results.