Personalized real-time movie recommendation system: Practical prototype and evaluation

Jiang Zhang, Yufeng Wang, Zhiyuan Yuan, Qun Jin

Research output: Contribution to journalArticle

Abstract

With the eruption of big data, practical recommendation schemes are now very important in various fields, including e-commerce, social networks, and a number of web-based services. Nowadays, there exist many personalized movie recommendation schemes utilizing publicly available movie datasets (e.g., MovieLens and Netflix), and returning improved performance metrics (e.g., Root-Mean-Square Error (RMSE)). However, two fundamental issues faced by movie recommendation systems are still neglected: first, scalability, and second, practical usage feedback and verification based on real implementation. In particular, Collaborative Filtering (CF) is one of the major prevailing techniques for implementing recommendation systems. However, traditional CF schemes suffer from a time complexity problem, which makes them bad candidates for real-world recommendation systems. In this paper, we address these two issues. Firstly, a simple but high-efficient recommendation algorithm is proposed, which exploits users' profile attributes to partition them into several clusters. For each cluster, a virtual opinion leader is conceived to represent the whole cluster, such that the dimension of the original useritem matrix can be significantly reduced, then a Weighted Slope One-VU method is designed and applied to the virtual opinion leader-item matrix to obtain the recommendation results. Compared to traditional clusteringbased CF recommendation schemes, our method can significantly reduce the time complexity, while achieving comparable recommendation performance. Furthermore, we have constructed a real personalized web-based movie recommendation system, MovieWatch, opened it to the public, collected user feedback on recommendations, and evaluated the feasibility and accuracy of our system based on this real-world data.

Original languageEnglish
Pages (from-to)180-191
Number of pages12
JournalTsinghua Science and Technology
Volume25
Issue number2
DOIs
Publication statusPublished - 2020 Apr 1

Fingerprint

Recommender systems
Collaborative filtering
Feedback
Mean square error
Scalability

Keywords

  • Collaborative filtering
  • Data mining
  • Movie recommendation system
  • Real-time
  • Virtual opinion leader

ASJC Scopus subject areas

  • General

Cite this

Personalized real-time movie recommendation system : Practical prototype and evaluation. / Zhang, Jiang; Wang, Yufeng; Yuan, Zhiyuan; Jin, Qun.

In: Tsinghua Science and Technology, Vol. 25, No. 2, 01.04.2020, p. 180-191.

Research output: Contribution to journalArticle

Zhang, Jiang ; Wang, Yufeng ; Yuan, Zhiyuan ; Jin, Qun. / Personalized real-time movie recommendation system : Practical prototype and evaluation. In: Tsinghua Science and Technology. 2020 ; Vol. 25, No. 2. pp. 180-191.
@article{1a12f1fdc71a41c98acab14e44e16919,
title = "Personalized real-time movie recommendation system: Practical prototype and evaluation",
abstract = "With the eruption of big data, practical recommendation schemes are now very important in various fields, including e-commerce, social networks, and a number of web-based services. Nowadays, there exist many personalized movie recommendation schemes utilizing publicly available movie datasets (e.g., MovieLens and Netflix), and returning improved performance metrics (e.g., Root-Mean-Square Error (RMSE)). However, two fundamental issues faced by movie recommendation systems are still neglected: first, scalability, and second, practical usage feedback and verification based on real implementation. In particular, Collaborative Filtering (CF) is one of the major prevailing techniques for implementing recommendation systems. However, traditional CF schemes suffer from a time complexity problem, which makes them bad candidates for real-world recommendation systems. In this paper, we address these two issues. Firstly, a simple but high-efficient recommendation algorithm is proposed, which exploits users' profile attributes to partition them into several clusters. For each cluster, a virtual opinion leader is conceived to represent the whole cluster, such that the dimension of the original useritem matrix can be significantly reduced, then a Weighted Slope One-VU method is designed and applied to the virtual opinion leader-item matrix to obtain the recommendation results. Compared to traditional clusteringbased CF recommendation schemes, our method can significantly reduce the time complexity, while achieving comparable recommendation performance. Furthermore, we have constructed a real personalized web-based movie recommendation system, MovieWatch, opened it to the public, collected user feedback on recommendations, and evaluated the feasibility and accuracy of our system based on this real-world data.",
keywords = "Collaborative filtering, Data mining, Movie recommendation system, Real-time, Virtual opinion leader",
author = "Jiang Zhang and Yufeng Wang and Zhiyuan Yuan and Qun Jin",
year = "2020",
month = "4",
day = "1",
doi = "10.26599/TST.2018.9010118",
language = "English",
volume = "25",
pages = "180--191",
journal = "Tsinghua Science and Technology",
issn = "1007-0214",
publisher = "Tsing Hua University",
number = "2",

}

TY - JOUR

T1 - Personalized real-time movie recommendation system

T2 - Practical prototype and evaluation

AU - Zhang, Jiang

AU - Wang, Yufeng

AU - Yuan, Zhiyuan

AU - Jin, Qun

PY - 2020/4/1

Y1 - 2020/4/1

N2 - With the eruption of big data, practical recommendation schemes are now very important in various fields, including e-commerce, social networks, and a number of web-based services. Nowadays, there exist many personalized movie recommendation schemes utilizing publicly available movie datasets (e.g., MovieLens and Netflix), and returning improved performance metrics (e.g., Root-Mean-Square Error (RMSE)). However, two fundamental issues faced by movie recommendation systems are still neglected: first, scalability, and second, practical usage feedback and verification based on real implementation. In particular, Collaborative Filtering (CF) is one of the major prevailing techniques for implementing recommendation systems. However, traditional CF schemes suffer from a time complexity problem, which makes them bad candidates for real-world recommendation systems. In this paper, we address these two issues. Firstly, a simple but high-efficient recommendation algorithm is proposed, which exploits users' profile attributes to partition them into several clusters. For each cluster, a virtual opinion leader is conceived to represent the whole cluster, such that the dimension of the original useritem matrix can be significantly reduced, then a Weighted Slope One-VU method is designed and applied to the virtual opinion leader-item matrix to obtain the recommendation results. Compared to traditional clusteringbased CF recommendation schemes, our method can significantly reduce the time complexity, while achieving comparable recommendation performance. Furthermore, we have constructed a real personalized web-based movie recommendation system, MovieWatch, opened it to the public, collected user feedback on recommendations, and evaluated the feasibility and accuracy of our system based on this real-world data.

AB - With the eruption of big data, practical recommendation schemes are now very important in various fields, including e-commerce, social networks, and a number of web-based services. Nowadays, there exist many personalized movie recommendation schemes utilizing publicly available movie datasets (e.g., MovieLens and Netflix), and returning improved performance metrics (e.g., Root-Mean-Square Error (RMSE)). However, two fundamental issues faced by movie recommendation systems are still neglected: first, scalability, and second, practical usage feedback and verification based on real implementation. In particular, Collaborative Filtering (CF) is one of the major prevailing techniques for implementing recommendation systems. However, traditional CF schemes suffer from a time complexity problem, which makes them bad candidates for real-world recommendation systems. In this paper, we address these two issues. Firstly, a simple but high-efficient recommendation algorithm is proposed, which exploits users' profile attributes to partition them into several clusters. For each cluster, a virtual opinion leader is conceived to represent the whole cluster, such that the dimension of the original useritem matrix can be significantly reduced, then a Weighted Slope One-VU method is designed and applied to the virtual opinion leader-item matrix to obtain the recommendation results. Compared to traditional clusteringbased CF recommendation schemes, our method can significantly reduce the time complexity, while achieving comparable recommendation performance. Furthermore, we have constructed a real personalized web-based movie recommendation system, MovieWatch, opened it to the public, collected user feedback on recommendations, and evaluated the feasibility and accuracy of our system based on this real-world data.

KW - Collaborative filtering

KW - Data mining

KW - Movie recommendation system

KW - Real-time

KW - Virtual opinion leader

UR - http://www.scopus.com/inward/record.url?scp=85072046151&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072046151&partnerID=8YFLogxK

U2 - 10.26599/TST.2018.9010118

DO - 10.26599/TST.2018.9010118

M3 - Article

AN - SCOPUS:85072046151

VL - 25

SP - 180

EP - 191

JO - Tsinghua Science and Technology

JF - Tsinghua Science and Technology

SN - 1007-0214

IS - 2

ER -