MATRIX FACTORIZATION TECHNIQUES FOR RECOMMENDER SYSTEMS

被引：7549

作者：

Koren, Yehuda ^{[1
]}

Bell, Robert ^{[1
]}

Volinsky, Chris ^{[1
]}

机构：

[1] Yahoo Res, Haifa, Israel

来源：

COMPUTER | 2009年 / 42卷 / 08期

关键词：

D O I：

10.1109/MC.2009.263

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

As the Netflix Prize competition has demonstrated, matrix factorization models are superior to classic nearest-neighbor techniques for producing product recommendations, allowing the incorporation of additional information such as implicit levels. © 2009, IEEE. All rights reserved.

引用

页码：30 / 37

页数：8

共 12 条

[1]

[Anonymous], 2009, Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining, DOI DOI 10.1145/1557019.1557072

[2]

[Anonymous], 2007, KDD CUP WORKSH

[3]

[Anonymous], 2007, SIGKDD Explorations, DOI DOI 10.1145/1345448.134546

[4]

BELLIVEAU PP, 2007, PHARM EDUC, V7, P43, DOI DOI 10.1080/15602210601084531

[5]

Funk Simon, 2006, NETFLIX UPDATE TRY T

[6] USING COLLABORATIVE FILTERING TO WEAVE AN INFORMATION TAPESTRY [J].

GOLDBERG, D ;

NICHOLS, D ;

OKI, BM ;

TERRY, D .

COMMUNICATIONS OF THE ACM, 1992, 35 (12) :61-70

[7] Collaborative Filtering for Implicit Feedback Datasets [J].

Hu, Yifan ;

Koren, Yehuda ;

Volinsky, Chris .

ICDM 2008: EIGHTH IEEE INTERNATIONAL CONFERENCE ON DATA MINING, PROCEEDINGS, 2008, :263-+

[8]

Koren Yehuda, 2008, PROC 14 ACM SIGKDD I, P426

[9]

Mnih A., 2008, Advances in Neural Information Processing Systems

[10]

Paterek A., 2007, P KDD CUP WORKSH4, P5, DOI [DOI 10.1145/1557019.1557072, DOI 10.1137/1.9781611972757.43]

← 1 2 →