Variational deep collaborative matrix factorization for social recommendation

Teng Xiao; Hui Tian; Hong Shen

doi:10.1007/978-3-030-16148-4_33

Variational deep collaborative matrix factorization for social recommendation

Teng Xiao, Hui Tian, Hong Shen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

6 Citations (Scopus)

Abstract

In this paper, we propose a Variational Deep Collaborative Matrix Factorization (VDCMF) algorithm for social recommendation that infers latent factors more effectively than existing methods by incorporating users’ social trust information and items’ content information into a unified generative framework. Unlike neural network-based algorithms, our model is not only effective in capturing the non-linearity among correlated variables but also powerful in predicting missing values under the robust collaborative inference. Specifically, we use variational auto-encoder to extract the latent representations of content and then incorporate them into traditional social trust factorization. We propose an efficient expectation-maximization inference algorithm to learn the model’s parameters and approximate the posteriors of latent factors. Experiments on two sparse datasets show that our VDCMF significantly outperforms major state-of-the-art CF methods for recommendation accuracy on common metrics.

Original language	English
Title of host publication	Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings
Editors	Qiang Yang, Zhiguo Gong, Zhi-Hua Zhou, Sheng-Jun Huang, Min-Ling Zhang
Publisher	Springer Verlag
Pages	426-437
Number of pages	12
ISBN (Print)	9783030161477
DOIs	https://doi.org/10.1007/978-3-030-16148-4_33
Publication status	Published - 2019
Externally published	Yes
Event	23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2019 - Macau, China Duration: 14 Apr 2019 → 17 Apr 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11439 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2019
Country/Territory	China
City	Macau
Period	14/04/19 → 17/04/19

Keywords

Deep Learning
Generative model
Matrix Factorization
Recommender System

Access to Document

10.1007/978-3-030-16148-4_33

Cite this

Xiao, T., Tian, H., & Shen, H. (2019). Variational deep collaborative matrix factorization for social recommendation. In Q. Yang, Z. Gong, Z.-H. Zhou, S.-J. Huang, & M.-L. Zhang (Eds.), Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings (pp. 426-437). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11439 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-16148-4_33

Xiao, Teng ; Tian, Hui ; Shen, Hong. / Variational deep collaborative matrix factorization for social recommendation. Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings. editor / Qiang Yang ; Zhiguo Gong ; Zhi-Hua Zhou ; Sheng-Jun Huang ; Min-Ling Zhang. Springer Verlag, 2019. pp. 426-437 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c58d273042a14e20addeded289c9c3b0,

title = "Variational deep collaborative matrix factorization for social recommendation",

abstract = "In this paper, we propose a Variational Deep Collaborative Matrix Factorization (VDCMF) algorithm for social recommendation that infers latent factors more effectively than existing methods by incorporating users{\textquoteright} social trust information and items{\textquoteright} content information into a unified generative framework. Unlike neural network-based algorithms, our model is not only effective in capturing the non-linearity among correlated variables but also powerful in predicting missing values under the robust collaborative inference. Specifically, we use variational auto-encoder to extract the latent representations of content and then incorporate them into traditional social trust factorization. We propose an efficient expectation-maximization inference algorithm to learn the model{\textquoteright}s parameters and approximate the posteriors of latent factors. Experiments on two sparse datasets show that our VDCMF significantly outperforms major state-of-the-art CF methods for recommendation accuracy on common metrics.",

keywords = "Deep Learning, Generative model, Matrix Factorization, Recommender System",

author = "Teng Xiao and Hui Tian and Hong Shen",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2019.; 23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2019 ; Conference date: 14-04-2019 Through 17-04-2019",

year = "2019",

doi = "10.1007/978-3-030-16148-4_33",

language = "English",

isbn = "9783030161477",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "426--437",

editor = "Qiang Yang and Zhiguo Gong and Zhi-Hua Zhou and Sheng-Jun Huang and Min-Ling Zhang",

booktitle = "Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings",

address = "Germany",

}

Xiao, T, Tian, H & Shen, H 2019, Variational deep collaborative matrix factorization for social recommendation. in Q Yang, Z Gong, Z-H Zhou, S-J Huang & M-L Zhang (eds), Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11439 LNAI, Springer Verlag, pp. 426-437, 23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2019, Macau, China, 14/04/19. https://doi.org/10.1007/978-3-030-16148-4_33

Variational deep collaborative matrix factorization for social recommendation. / Xiao, Teng; Tian, Hui; Shen, Hong.
Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings. ed. / Qiang Yang; Zhiguo Gong; Zhi-Hua Zhou; Sheng-Jun Huang; Min-Ling Zhang. Springer Verlag, 2019. p. 426-437 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11439 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Variational deep collaborative matrix factorization for social recommendation

AU - Xiao, Teng

AU - Tian, Hui

AU - Shen, Hong

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2019.

PY - 2019

Y1 - 2019

N2 - In this paper, we propose a Variational Deep Collaborative Matrix Factorization (VDCMF) algorithm for social recommendation that infers latent factors more effectively than existing methods by incorporating users’ social trust information and items’ content information into a unified generative framework. Unlike neural network-based algorithms, our model is not only effective in capturing the non-linearity among correlated variables but also powerful in predicting missing values under the robust collaborative inference. Specifically, we use variational auto-encoder to extract the latent representations of content and then incorporate them into traditional social trust factorization. We propose an efficient expectation-maximization inference algorithm to learn the model’s parameters and approximate the posteriors of latent factors. Experiments on two sparse datasets show that our VDCMF significantly outperforms major state-of-the-art CF methods for recommendation accuracy on common metrics.

AB - In this paper, we propose a Variational Deep Collaborative Matrix Factorization (VDCMF) algorithm for social recommendation that infers latent factors more effectively than existing methods by incorporating users’ social trust information and items’ content information into a unified generative framework. Unlike neural network-based algorithms, our model is not only effective in capturing the non-linearity among correlated variables but also powerful in predicting missing values under the robust collaborative inference. Specifically, we use variational auto-encoder to extract the latent representations of content and then incorporate them into traditional social trust factorization. We propose an efficient expectation-maximization inference algorithm to learn the model’s parameters and approximate the posteriors of latent factors. Experiments on two sparse datasets show that our VDCMF significantly outperforms major state-of-the-art CF methods for recommendation accuracy on common metrics.

KW - Deep Learning

KW - Generative model

KW - Matrix Factorization

KW - Recommender System

UR - http://www.scopus.com/inward/record.url?scp=85064920147&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-16148-4_33

DO - 10.1007/978-3-030-16148-4_33

M3 - Conference contribution

AN - SCOPUS:85064920147

SN - 9783030161477

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 426

EP - 437

BT - Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings

A2 - Yang, Qiang

A2 - Gong, Zhiguo

A2 - Zhou, Zhi-Hua

A2 - Huang, Sheng-Jun

A2 - Zhang, Min-Ling

PB - Springer Verlag

T2 - 23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2019

Y2 - 14 April 2019 through 17 April 2019

ER -

Xiao T, Tian H, Shen H. Variational deep collaborative matrix factorization for social recommendation. In Yang Q, Gong Z, Zhou ZH, Huang SJ, Zhang ML, editors, Advances in Knowledge Discovery and Data Mining - 23rd Pacific-Asia Conference, PAKDD 2019, Proceedings. Springer Verlag. 2019. p. 426-437. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-16148-4_33

Variational deep collaborative matrix factorization for social recommendation

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this