GPS: Factorized group preference-based similarity models for sparse sequential recommendation

Yeongwook Yang, Danial Hooshyar, Heui Seok Lim

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

One of the key tasks for recommender systems is the prediction of personalized sequential behavior. There are two primary means of modeling sequential patterns and long-term user preferences: Markov chains and matrix factorization, respectively. Together, they provide a unified approach to predicting user actions. In spite of their strengths in tackling dense data, however, these methods struggle with the sparsity issues often present in real-world datasets. In approaching this problem, we propose combining similarity-based methods (demonstrably helpful for sequentially unaware item recommendation) with Markov chains to offer individualized sequential recommendations. This approach, called GPS (a factorized group preference-based similarity model), further leverages the idea of group preference along with user preference to introduce a greater array of interactions between users—which in turn eases the problem of data sparsity and cold users and cuts down on the assumption of a strong independency within various factors. By applying our method to a range of large, real-world datasets, we demonstrate quantitatively that GPS outperforms several state-of-the-art methods, particularly in cases with sparse datasets. Regarding qualitative findings, GPS also grasps personalized interactions and can provide recommendations that are both on-target and meaningful.

Original languageEnglish
Pages (from-to)394-411
Number of pages18
JournalInformation Sciences
Volume481
DOIs
Publication statusPublished - 2019 May 1

Fingerprint

Global positioning system
Recommendations
Markov processes
User Preferences
Sparsity
Markov chain
Recommender systems
Factorization
Sequential Patterns
Matrix Factorization
Recommender Systems
Interaction
Leverage
Model
Target
Similarity
Prediction
Modeling
Range of data
Demonstrate

Keywords

  • Group preference
  • Recommender systems
  • Sequential recommendation
  • Similarity models

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Theoretical Computer Science
  • Computer Science Applications
  • Information Systems and Management
  • Artificial Intelligence

Cite this

GPS : Factorized group preference-based similarity models for sparse sequential recommendation. / Yang, Yeongwook; Hooshyar, Danial; Lim, Heui Seok.

In: Information Sciences, Vol. 481, 01.05.2019, p. 394-411.

Research output: Contribution to journalArticle

@article{628e3cb459ca4d5d90bacfbf01e1c60c,
title = "GPS: Factorized group preference-based similarity models for sparse sequential recommendation",
abstract = "One of the key tasks for recommender systems is the prediction of personalized sequential behavior. There are two primary means of modeling sequential patterns and long-term user preferences: Markov chains and matrix factorization, respectively. Together, they provide a unified approach to predicting user actions. In spite of their strengths in tackling dense data, however, these methods struggle with the sparsity issues often present in real-world datasets. In approaching this problem, we propose combining similarity-based methods (demonstrably helpful for sequentially unaware item recommendation) with Markov chains to offer individualized sequential recommendations. This approach, called GPS (a factorized group preference-based similarity model), further leverages the idea of group preference along with user preference to introduce a greater array of interactions between users—which in turn eases the problem of data sparsity and cold users and cuts down on the assumption of a strong independency within various factors. By applying our method to a range of large, real-world datasets, we demonstrate quantitatively that GPS outperforms several state-of-the-art methods, particularly in cases with sparse datasets. Regarding qualitative findings, GPS also grasps personalized interactions and can provide recommendations that are both on-target and meaningful.",
keywords = "Group preference, Recommender systems, Sequential recommendation, Similarity models",
author = "Yeongwook Yang and Danial Hooshyar and Lim, {Heui Seok}",
year = "2019",
month = "5",
day = "1",
doi = "10.1016/j.ins.2018.12.053",
language = "English",
volume = "481",
pages = "394--411",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - GPS

T2 - Factorized group preference-based similarity models for sparse sequential recommendation

AU - Yang, Yeongwook

AU - Hooshyar, Danial

AU - Lim, Heui Seok

PY - 2019/5/1

Y1 - 2019/5/1

N2 - One of the key tasks for recommender systems is the prediction of personalized sequential behavior. There are two primary means of modeling sequential patterns and long-term user preferences: Markov chains and matrix factorization, respectively. Together, they provide a unified approach to predicting user actions. In spite of their strengths in tackling dense data, however, these methods struggle with the sparsity issues often present in real-world datasets. In approaching this problem, we propose combining similarity-based methods (demonstrably helpful for sequentially unaware item recommendation) with Markov chains to offer individualized sequential recommendations. This approach, called GPS (a factorized group preference-based similarity model), further leverages the idea of group preference along with user preference to introduce a greater array of interactions between users—which in turn eases the problem of data sparsity and cold users and cuts down on the assumption of a strong independency within various factors. By applying our method to a range of large, real-world datasets, we demonstrate quantitatively that GPS outperforms several state-of-the-art methods, particularly in cases with sparse datasets. Regarding qualitative findings, GPS also grasps personalized interactions and can provide recommendations that are both on-target and meaningful.

AB - One of the key tasks for recommender systems is the prediction of personalized sequential behavior. There are two primary means of modeling sequential patterns and long-term user preferences: Markov chains and matrix factorization, respectively. Together, they provide a unified approach to predicting user actions. In spite of their strengths in tackling dense data, however, these methods struggle with the sparsity issues often present in real-world datasets. In approaching this problem, we propose combining similarity-based methods (demonstrably helpful for sequentially unaware item recommendation) with Markov chains to offer individualized sequential recommendations. This approach, called GPS (a factorized group preference-based similarity model), further leverages the idea of group preference along with user preference to introduce a greater array of interactions between users—which in turn eases the problem of data sparsity and cold users and cuts down on the assumption of a strong independency within various factors. By applying our method to a range of large, real-world datasets, we demonstrate quantitatively that GPS outperforms several state-of-the-art methods, particularly in cases with sparse datasets. Regarding qualitative findings, GPS also grasps personalized interactions and can provide recommendations that are both on-target and meaningful.

KW - Group preference

KW - Recommender systems

KW - Sequential recommendation

KW - Similarity models

UR - http://www.scopus.com/inward/record.url?scp=85059581276&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059581276&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2018.12.053

DO - 10.1016/j.ins.2018.12.053

M3 - Article

AN - SCOPUS:85059581276

VL - 481

SP - 394

EP - 411

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -