Bayesian latent variable models for collaborative item rating prediction

Morgan Alexander Harvey, M. Carman, Ian Ruthven, Fabio Crestani

Research output: Contribution to conferencePaper

28 Citations (Scopus)

Abstract

Collaborative filtering systems based on ratings make it easier for users to find content of interest on the Web and as such they constitute an area of much research. In this paper we first present a Bayesian latent variable model for rating prediction that models ratings over each user's latent interests and also each item's latent topics. We describe a Gibbs sampling procedure that can be used to estimate its parameters and show by experiment that it is competitive with the gradient descent SVD methods commonly used in state-of-the-art systems. We then proceed to make an important and novel extension to this model, enhancing it with user-dependent and item-dependant biases to significantly improve rating estimation. We show by experiment on a large set of real ratings data that these models are able to outperform 3 common baselines, including a very competitive and modern SVD-based model. Furthermore we illustrate other advantages of our approach beyond simply its ability to provide more accurate ratings and show that it is able to perform better on the common and important case where the user profile is short.
LanguageEnglish
Pages699-708
Number of pages10
DOIs
Publication statusPublished - Oct 2011
EventACM CIKM -
Duration: 31 Mar 2011 → …

Other

OtherACM CIKM
Period31/03/11 → …

Fingerprint

Singular value decomposition
Collaborative filtering
Experiments
Sampling

Keywords

  • Bayesian latent variable models
  • collaborative item rating prediction

Cite this

Harvey, M. A., Carman, M., Ruthven, I., & Crestani, F. (2011). Bayesian latent variable models for collaborative item rating prediction. 699-708. Paper presented at ACM CIKM, . https://doi.org/10.1145/2063576.2063680
Harvey, Morgan Alexander ; Carman, M. ; Ruthven, Ian ; Crestani, Fabio. / Bayesian latent variable models for collaborative item rating prediction. Paper presented at ACM CIKM, .10 p.
@conference{c61b66991a1c475cb793863f34933cb8,
title = "Bayesian latent variable models for collaborative item rating prediction",
abstract = "Collaborative filtering systems based on ratings make it easier for users to find content of interest on the Web and as such they constitute an area of much research. In this paper we first present a Bayesian latent variable model for rating prediction that models ratings over each user's latent interests and also each item's latent topics. We describe a Gibbs sampling procedure that can be used to estimate its parameters and show by experiment that it is competitive with the gradient descent SVD methods commonly used in state-of-the-art systems. We then proceed to make an important and novel extension to this model, enhancing it with user-dependent and item-dependant biases to significantly improve rating estimation. We show by experiment on a large set of real ratings data that these models are able to outperform 3 common baselines, including a very competitive and modern SVD-based model. Furthermore we illustrate other advantages of our approach beyond simply its ability to provide more accurate ratings and show that it is able to perform better on the common and important case where the user profile is short.",
keywords = "Bayesian latent variable models , collaborative item rating prediction",
author = "Harvey, {Morgan Alexander} and M. Carman and Ian Ruthven and Fabio Crestani",
year = "2011",
month = "10",
doi = "10.1145/2063576.2063680",
language = "English",
pages = "699--708",
note = "ACM CIKM ; Conference date: 31-03-2011",

}

Harvey, MA, Carman, M, Ruthven, I & Crestani, F 2011, 'Bayesian latent variable models for collaborative item rating prediction' Paper presented at ACM CIKM, 31/03/11, pp. 699-708. https://doi.org/10.1145/2063576.2063680

Bayesian latent variable models for collaborative item rating prediction. / Harvey, Morgan Alexander; Carman, M.; Ruthven, Ian; Crestani, Fabio.

2011. 699-708 Paper presented at ACM CIKM, .

Research output: Contribution to conferencePaper

TY - CONF

T1 - Bayesian latent variable models for collaborative item rating prediction

AU - Harvey, Morgan Alexander

AU - Carman, M.

AU - Ruthven, Ian

AU - Crestani, Fabio

PY - 2011/10

Y1 - 2011/10

N2 - Collaborative filtering systems based on ratings make it easier for users to find content of interest on the Web and as such they constitute an area of much research. In this paper we first present a Bayesian latent variable model for rating prediction that models ratings over each user's latent interests and also each item's latent topics. We describe a Gibbs sampling procedure that can be used to estimate its parameters and show by experiment that it is competitive with the gradient descent SVD methods commonly used in state-of-the-art systems. We then proceed to make an important and novel extension to this model, enhancing it with user-dependent and item-dependant biases to significantly improve rating estimation. We show by experiment on a large set of real ratings data that these models are able to outperform 3 common baselines, including a very competitive and modern SVD-based model. Furthermore we illustrate other advantages of our approach beyond simply its ability to provide more accurate ratings and show that it is able to perform better on the common and important case where the user profile is short.

AB - Collaborative filtering systems based on ratings make it easier for users to find content of interest on the Web and as such they constitute an area of much research. In this paper we first present a Bayesian latent variable model for rating prediction that models ratings over each user's latent interests and also each item's latent topics. We describe a Gibbs sampling procedure that can be used to estimate its parameters and show by experiment that it is competitive with the gradient descent SVD methods commonly used in state-of-the-art systems. We then proceed to make an important and novel extension to this model, enhancing it with user-dependent and item-dependant biases to significantly improve rating estimation. We show by experiment on a large set of real ratings data that these models are able to outperform 3 common baselines, including a very competitive and modern SVD-based model. Furthermore we illustrate other advantages of our approach beyond simply its ability to provide more accurate ratings and show that it is able to perform better on the common and important case where the user profile is short.

KW - Bayesian latent variable models

KW - collaborative item rating prediction

UR - http://dl.acm.org/citation.cfm?id=2063680

U2 - 10.1145/2063576.2063680

DO - 10.1145/2063576.2063680

M3 - Paper

SP - 699

EP - 708

ER -