Matrix Factorization with Rating Completion: an Enhanced SVD Model for Collaborative Filtering Recommender Systems

Xin Guan, Chang Tsun Li, Yu Guan.

Research output: Contribution to journalArticle

  • 6 Citations

Abstract

Collaborative filtering algorithms such as matrix factorization techniques are recently gaining momentum due to their promising performance on recommender systems. However, most collaborative filtering algorithms suffer from data sparsity. Active learning algorithms are effective in reducing the sparsity problem for recommender systems by requesting users to give ratings to some items when they enter the systems. In this paper, a new matrix factorization model, called Enhanced SVD (ESVD) is proposed, which incorporates the classic matrix factorization algorithms with ratings completion inspired by active learning. In addition, the connection between the prediction accuracy and the density of matrix is built to further explore its potentials. We also propose the Multi-layer ESVD (MESVD), which learns the model iteratively to further improve the prediction accuracy. To handle the imbalanced datasets that contain far more users than items or more items than users, the Item-wise ESVD (IESVD) and User-wise ESVD (UESVD) are presented, respectively. The proposed methods are evaluated on the famous Netflix and Movielens datasets. Experimental results validate their effectiveness in terms of both accuracy and efficiency when compared with traditional matrix factorization methods and active learning methods.

LanguageEnglish
Article number27668
Pages 27668 - 27678
Number of pages11
JournalIEEE Access
Volume5
DOIs
Publication statusPublished - 22 Dec 2017

Fingerprint

Collaborative filtering
Recommender systems
Singular value decomposition
Factorization
Learning algorithms
Momentum
Problem-Based Learning

Cite this

@article{eb764517bced45db81f5ba13d3063108,
title = "Matrix Factorization with Rating Completion: an Enhanced SVD Model for Collaborative Filtering Recommender Systems",
abstract = "Collaborative filtering algorithms such as matrix factorization techniques are recently gaining momentum due to their promising performance on recommender systems. However, most collaborative filtering algorithms suffer from data sparsity. Active learning algorithms are effective in reducing the sparsity problem for recommender systems by requesting users to give ratings to some items when they enter the systems. In this paper, a new matrix factorization model, called Enhanced SVD (ESVD) is proposed, which incorporates the classic matrix factorization algorithms with ratings completion inspired by active learning. In addition, the connection between the prediction accuracy and the density of matrix is built to further explore its potentials. We also propose the Multi-layer ESVD (MESVD), which learns the model iteratively to further improve the prediction accuracy. To handle the imbalanced datasets that contain far more users than items or more items than users, the Item-wise ESVD (IESVD) and User-wise ESVD (UESVD) are presented, respectively. The proposed methods are evaluated on the famous Netflix and Movielens datasets. Experimental results validate their effectiveness in terms of both accuracy and efficiency when compared with traditional matrix factorization methods and active learning methods.",
keywords = "active learning, Collaboration, Computational modeling, data sparseness, matrix factorization, Motion pictures, Prediction algorithms, Predictive models, rating completion, Recommender systems, recommender systems, Sparse matrices",
author = "Xin Guan and Li, {Chang Tsun} and Yu Guan.",
year = "2017",
month = "12",
day = "22",
doi = "10.1109/ACCESS.2017.2772226",
language = "English",
volume = "5",
pages = "27668 -- 27678",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",

}

Matrix Factorization with Rating Completion : an Enhanced SVD Model for Collaborative Filtering Recommender Systems. / Guan, Xin; Li, Chang Tsun; Guan., Yu.

In: IEEE Access, Vol. 5, 27668, 22.12.2017, p. 27668 - 27678.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Matrix Factorization with Rating Completion

T2 - IEEE Access

AU - Guan, Xin

AU - Li, Chang Tsun

AU - Guan., Yu

PY - 2017/12/22

Y1 - 2017/12/22

N2 - Collaborative filtering algorithms such as matrix factorization techniques are recently gaining momentum due to their promising performance on recommender systems. However, most collaborative filtering algorithms suffer from data sparsity. Active learning algorithms are effective in reducing the sparsity problem for recommender systems by requesting users to give ratings to some items when they enter the systems. In this paper, a new matrix factorization model, called Enhanced SVD (ESVD) is proposed, which incorporates the classic matrix factorization algorithms with ratings completion inspired by active learning. In addition, the connection between the prediction accuracy and the density of matrix is built to further explore its potentials. We also propose the Multi-layer ESVD (MESVD), which learns the model iteratively to further improve the prediction accuracy. To handle the imbalanced datasets that contain far more users than items or more items than users, the Item-wise ESVD (IESVD) and User-wise ESVD (UESVD) are presented, respectively. The proposed methods are evaluated on the famous Netflix and Movielens datasets. Experimental results validate their effectiveness in terms of both accuracy and efficiency when compared with traditional matrix factorization methods and active learning methods.

AB - Collaborative filtering algorithms such as matrix factorization techniques are recently gaining momentum due to their promising performance on recommender systems. However, most collaborative filtering algorithms suffer from data sparsity. Active learning algorithms are effective in reducing the sparsity problem for recommender systems by requesting users to give ratings to some items when they enter the systems. In this paper, a new matrix factorization model, called Enhanced SVD (ESVD) is proposed, which incorporates the classic matrix factorization algorithms with ratings completion inspired by active learning. In addition, the connection between the prediction accuracy and the density of matrix is built to further explore its potentials. We also propose the Multi-layer ESVD (MESVD), which learns the model iteratively to further improve the prediction accuracy. To handle the imbalanced datasets that contain far more users than items or more items than users, the Item-wise ESVD (IESVD) and User-wise ESVD (UESVD) are presented, respectively. The proposed methods are evaluated on the famous Netflix and Movielens datasets. Experimental results validate their effectiveness in terms of both accuracy and efficiency when compared with traditional matrix factorization methods and active learning methods.

KW - active learning

KW - Collaboration

KW - Computational modeling

KW - data sparseness

KW - matrix factorization

KW - Motion pictures

KW - Prediction algorithms

KW - Predictive models

KW - rating completion

KW - Recommender systems

KW - recommender systems

KW - Sparse matrices

UR - http://www.scopus.com/inward/record.url?scp=85035791015&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85035791015&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2017.2772226

DO - 10.1109/ACCESS.2017.2772226

M3 - Article

VL - 5

SP - 27668

EP - 27678

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 27668

ER -