Abstract:Recommender systems are essential tools for many e-commerce services, such as Amazon, Netflix, etc. to recommend new items to users. Among various recommendation techniques, collaborative filtering has shown tremendous performance by using the rating patterns of users. Traditional collaborative filtering, matrix factorization, and deep matrix factorization are the most representative collaborative filtering techniques. However, despite their extensive utility, the selection of the method which provides better recommendation performance is still a major concern in multi-criteria recommender systems. Most recommender systems (RSs) work only on the single criterion rating, i.e., the overall rating. Single-criterion collaborative filtering (CF) generates less reliable recommendations because it suffers from correlation-based similarity problems. However, predictions based on multiple criteria have proven more accurate. This paper compares traditional collaborative filtering, matrix factorization and deep matrix factorization in recommender systems on multi-criteria datasets. We describe details of these techniques in various aspects of recommendation quality, such as how those methods handle cold-start problems, which typically happen in collaborative filtering. We performed several experiments extensively over two real-world datasets to evaluate the performance of each method in terms of qualitative and quantitative measures and observe that deep matrix factorization techniques is superior to all other techniques.

An Item-Based Collaborative Filtering using Dimensionality Reduction Techniques on Mahout Framework

An Effective Algorithm For Dimensional Reduction In Collaborative Filtering

A Collaborative Filtering-Based Recommender Systems approach for Multifarious applications

Information Filtering Via Collaborative User Clustering Modeling.

An Improved Collaborative Filtering Algorithm based on Dimension Reduction and Improved Clustering

A hybrid knowledge-based approach to collaborative filtering for improved recommendations

DESIGN AND ANALYSIS OF COLLABORATIVE FILTERING BASED RECOMMENDATION SYSTEM

Multi-dimensional Adaptive Collaborative Filtering Recommendation Algorithm

Incorporating Community Detection and Clustering Techniques into Collaborative Filtering Model

An Efficient Collaborative Filtering Approach Using Smoothing and Fusing

Mining Intrinsic Information by Matrix Factorization-Based Approaches for Collaborative Filtering in Recommender Systems

Multi-Domain Collaborative Filtering

Deep Collaborative Filtering Incorporating Auxiliary Multi-Media Information.

Deep Learning Based Matrix Factorization for Collaborative Filtering

Comparative analysis of collaborative filtering techniques for the multi-criteria recommender systems

Distributed-Representation Based Hybrid Recommender System with Short Item Descriptions.

Integrating user-side information into matrix factorization to address data sparsity of collaborative filtering

A new collaborative filtering recommendation algorithm based on dimensionality reduction and clustering techniques

Multi-Task Matrix Factorization For Collaborative Filtering

Unified Matrix Factorization with Dynamic Multi-view Clustering

Cbmr: an Optimized Mapreduce for Item-Based Collaborative Filtering Recommendation Algorithm with Empirical Analysis