Recommender System Using Model Based Filtering

Introduction

With the development of artificial intelligence technology, increasingly more intelligent products are being applied in daily life and provide convenience for people in various aspects. The recommendation algorithm is the most important part of a recommendation system and directly determines the quality of the recommendation results and the performance of the system. The commonly used algorithms can be divided into two main categories: content-based filtering methods and collaborative filtering methods.

Content-based filtering methods construct portraits of users and items through the analysis of extra information, such as document content, user profiles, and the attributes of items, to make recommendations. In most cases, the information that is used to construct the portraits is difficult to obtain or even fake; therefore, its performance and application range suffer from significant limitations.

Collaborative filtering algorithms are the most widely used algorithms in recommendation systems; they are different from content-based methods in that they do not require information about users or items, and they make accurate recommendations based only on interaction information between users and items such as clicks, browsing, and rating. Although this method is simple and effective, with the rapid development of the Internet, the high sparsity of the data limits the performance of the algorithm; therefore, researchers have begun to look for other methods of improving the recommendation performance.

A third and more popular approach is Modal based filtering algorithm. The model-based methods are developed with data mining and machine learning algorithms to find patterns based on training data. Popular techniques of the model-based methods include the Bayesian models, clustering models, latent semantic models such as singular value decomposition, and probabilistic latent semantic analysis. The technique we will going to use is single value decomposition based on a popular algorithm known as Low-Rank Matrix Factorisation. A statistical approach to discover hidden correlation/ features in raw data.