Matrix factorization and neighbor based algorithms for the netflix prize problem

Gábor Takács, István Pilászy, Bottyán Németh, D. Tikk

Research output: Chapter in Book/Report/Conference proceedingConference contribution

92 Citations (Scopus)

Abstract

Collaborative filtering (CF) approaches proved to be effective for recommender systems in predicting user preferences in item selection using known user ratings of items. This subfield of machine learning has gained a lot of popularity with the Netix Prize competition started in October 2006. Two major approaches for this problem are matrix factorization (MF) and the neighbor based approach (NB). In this work, we propose various variants of MF and NB that can boost the performance of the usual ensemble based scheme. First, we investigate various regularization scenarios for MF. Second, we introduce two NB methods: one is based on correlation coeficients and the other on linear least squares. At the experimentation part, we show that the proposed approaches compare favorably with existing ones in terms of prediction accuracy and/or required training time. We present results of blending the proposed methods.

Original languageEnglish
Title of host publicationRecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems
Pages267-274
Number of pages8
DOIs
Publication statusPublished - 2008
Event2008 2nd ACM International Conference on Recommender Systems, RecSys'08 - Lausanne, Switzerland
Duration: Oct 23 2008Oct 25 2008

Other

Other2008 2nd ACM International Conference on Recommender Systems, RecSys'08
CountrySwitzerland
CityLausanne
Period10/23/0810/25/08

Fingerprint

Factorization
Collaborative filtering
Recommender systems
Learning systems

Keywords

  • Collaborative filtering
  • Matrix factorization
  • Neighbor based methods
  • Netix prize

ASJC Scopus subject areas

  • Information Systems
  • Control and Systems Engineering

Cite this

Takács, G., Pilászy, I., Németh, B., & Tikk, D. (2008). Matrix factorization and neighbor based algorithms for the netflix prize problem. In RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems (pp. 267-274) https://doi.org/10.1145/1454008.1454049

Matrix factorization and neighbor based algorithms for the netflix prize problem. / Takács, Gábor; Pilászy, István; Németh, Bottyán; Tikk, D.

RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems. 2008. p. 267-274.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Takács, G, Pilászy, I, Németh, B & Tikk, D 2008, Matrix factorization and neighbor based algorithms for the netflix prize problem. in RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems. pp. 267-274, 2008 2nd ACM International Conference on Recommender Systems, RecSys'08, Lausanne, Switzerland, 10/23/08. https://doi.org/10.1145/1454008.1454049
Takács G, Pilászy I, Németh B, Tikk D. Matrix factorization and neighbor based algorithms for the netflix prize problem. In RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems. 2008. p. 267-274 https://doi.org/10.1145/1454008.1454049
Takács, Gábor ; Pilászy, István ; Németh, Bottyán ; Tikk, D. / Matrix factorization and neighbor based algorithms for the netflix prize problem. RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems. 2008. pp. 267-274
@inproceedings{678cbc1a62024e268d196a23b6865800,
title = "Matrix factorization and neighbor based algorithms for the netflix prize problem",
abstract = "Collaborative filtering (CF) approaches proved to be effective for recommender systems in predicting user preferences in item selection using known user ratings of items. This subfield of machine learning has gained a lot of popularity with the Netix Prize competition started in October 2006. Two major approaches for this problem are matrix factorization (MF) and the neighbor based approach (NB). In this work, we propose various variants of MF and NB that can boost the performance of the usual ensemble based scheme. First, we investigate various regularization scenarios for MF. Second, we introduce two NB methods: one is based on correlation coeficients and the other on linear least squares. At the experimentation part, we show that the proposed approaches compare favorably with existing ones in terms of prediction accuracy and/or required training time. We present results of blending the proposed methods.",
keywords = "Collaborative filtering, Matrix factorization, Neighbor based methods, Netix prize",
author = "G{\'a}bor Tak{\'a}cs and Istv{\'a}n Pil{\'a}szy and Botty{\'a}n N{\'e}meth and D. Tikk",
year = "2008",
doi = "10.1145/1454008.1454049",
language = "English",
isbn = "9781605580937",
pages = "267--274",
booktitle = "RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems",

}

TY - GEN

T1 - Matrix factorization and neighbor based algorithms for the netflix prize problem

AU - Takács, Gábor

AU - Pilászy, István

AU - Németh, Bottyán

AU - Tikk, D.

PY - 2008

Y1 - 2008

N2 - Collaborative filtering (CF) approaches proved to be effective for recommender systems in predicting user preferences in item selection using known user ratings of items. This subfield of machine learning has gained a lot of popularity with the Netix Prize competition started in October 2006. Two major approaches for this problem are matrix factorization (MF) and the neighbor based approach (NB). In this work, we propose various variants of MF and NB that can boost the performance of the usual ensemble based scheme. First, we investigate various regularization scenarios for MF. Second, we introduce two NB methods: one is based on correlation coeficients and the other on linear least squares. At the experimentation part, we show that the proposed approaches compare favorably with existing ones in terms of prediction accuracy and/or required training time. We present results of blending the proposed methods.

AB - Collaborative filtering (CF) approaches proved to be effective for recommender systems in predicting user preferences in item selection using known user ratings of items. This subfield of machine learning has gained a lot of popularity with the Netix Prize competition started in October 2006. Two major approaches for this problem are matrix factorization (MF) and the neighbor based approach (NB). In this work, we propose various variants of MF and NB that can boost the performance of the usual ensemble based scheme. First, we investigate various regularization scenarios for MF. Second, we introduce two NB methods: one is based on correlation coeficients and the other on linear least squares. At the experimentation part, we show that the proposed approaches compare favorably with existing ones in terms of prediction accuracy and/or required training time. We present results of blending the proposed methods.

KW - Collaborative filtering

KW - Matrix factorization

KW - Neighbor based methods

KW - Netix prize

UR - http://www.scopus.com/inward/record.url?scp=63449123891&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=63449123891&partnerID=8YFLogxK

U2 - 10.1145/1454008.1454049

DO - 10.1145/1454008.1454049

M3 - Conference contribution

AN - SCOPUS:63449123891

SN - 9781605580937

SP - 267

EP - 274

BT - RecSys'08: Proceedings of the 2008 ACM Conference on Recommender Systems

ER -