Collaborative and content-based filtering are two paradigms that have been applied in the context of recommender systems and user preference prediction. This paper proposes a novel, unified approach that systematically integrates all available training information such as past user-item ratings as well as attributes of items or users to learn a prediction function. The key ingredient of our method is the design of a suitable kernel or similarity function between user-item pairs that allows simultaneous generalization across the user and item dimensions. We propose an on-line algorithm (JRank) that generalizes perceptron learning. Experimental results on the EachMovie data set show significant improvements over standard approaches.
| Author(s): | Basilico, J. and Hofmann, T. |
| Book Title: | ACM International Conference Proceeding Series |
| Journal: | Proceedings of the 21st International Conference on Machine Learning |
| Pages: | 65 |
| Year: | 2004 |
| Day: | 0 |
| Editors: | Greiner, R. , D. Schuurmans |
| Publisher: | ACM Press |
| Bibtex Type: | Conference Paper (inproceedings) |
| Address: | New York, USA |
| Event Name: | ICLM 2004 |
| Event Place: | Banff, Alberta, Canada |
| Digital: | 0 |
| Electronic Archiving: | grant_archive |
| Institution: | Max-Planck for biological Cybernetics, Tübingen, Germany |
| Organization: | Max-Planck-Gesellschaft |
| School: | Biologische Kybernetik |
| Links: | |
BibTex
@inproceedings{2739,
title = {Unifying Colloborative and Content-Based Filtering.},
journal = {Proceedings of the 21st International Conference on Machine Learning},
booktitle = {ACM International Conference Proceeding Series},
abstract = {Collaborative and content-based filtering are two paradigms that have been applied in the context of recommender systems and user
preference prediction. This paper proposes a novel, unified approach that systematically integrates all available training information
such as past user-item ratings as well as attributes of items or users to learn a prediction
function. The key ingredient of our method is the design of a suitable kernel or similarity function between user-item pairs that allows
simultaneous generalization across the user and item dimensions. We propose an on-line algorithm (JRank) that generalizes perceptron
learning. Experimental results on the EachMovie data set show significant improvements over standard approaches.},
pages = {65 },
editors = {Greiner, R. , D. Schuurmans},
publisher = {ACM Press},
organization = {Max-Planck-Gesellschaft},
institution = {Max-Planck for biological Cybernetics, Tübingen, Germany},
school = {Biologische Kybernetik},
address = {New York, USA},
year = {2004},
slug = {2739},
author = {Basilico, J. and Hofmann, T.}
}