Recommender systems traditionally assume that user profiles and movie attributes are static. Temporal dynamics are purely reactive, that is, they are inferred after they are observed, e.g. after a user's taste has changed or based on hand-engineered temporal bias corrections for movies. We propose Recurrent Recommender Networks (RRN) that are able to predict future behavioral trajectories. This is achieved by endowing both users and movies with a Long Short-Term Memory (LSTM) autoregressive model that captures dynamics, in addition to a more traditional low-rank factorization. On multiple real-world datasets, our model offers excellent prediction accuracy and it is very compact, since we need not learn latent state but rather just the state transition function.
%0 Conference Paper
%1 wu2017recurrent
%A Wu, Chao-Yuan
%A Ahmed, Amr
%A Beutel, Alex
%A Smola, Alexander J.
%A Jing, How
%B Proceedings of the Tenth ACM International Conference on Web Search and Data Mining
%C New York, NY, USA
%D 2017
%I ACM
%K recommendation rnn sys:toread
%P 495--503
%R 10.1145/3018661.3018689
%T Recurrent Recommender Networks
%U http://doi.acm.org/10.1145/3018661.3018689
%X Recommender systems traditionally assume that user profiles and movie attributes are static. Temporal dynamics are purely reactive, that is, they are inferred after they are observed, e.g. after a user's taste has changed or based on hand-engineered temporal bias corrections for movies. We propose Recurrent Recommender Networks (RRN) that are able to predict future behavioral trajectories. This is achieved by endowing both users and movies with a Long Short-Term Memory (LSTM) autoregressive model that captures dynamics, in addition to a more traditional low-rank factorization. On multiple real-world datasets, our model offers excellent prediction accuracy and it is very compact, since we need not learn latent state but rather just the state transition function.
%@ 978-1-4503-4675-7
@inproceedings{wu2017recurrent,
abstract = {Recommender systems traditionally assume that user profiles and movie attributes are static. Temporal dynamics are purely reactive, that is, they are inferred after they are observed, e.g. after a user's taste has changed or based on hand-engineered temporal bias corrections for movies. We propose Recurrent Recommender Networks (RRN) that are able to predict future behavioral trajectories. This is achieved by endowing both users and movies with a Long Short-Term Memory (LSTM) autoregressive model that captures dynamics, in addition to a more traditional low-rank factorization. On multiple real-world datasets, our model offers excellent prediction accuracy and it is very compact, since we need not learn latent state but rather just the state transition function.},
acmid = {3018689},
added-at = {2017-02-26T18:10:02.000+0100},
address = {New York, NY, USA},
author = {Wu, Chao-Yuan and Ahmed, Amr and Beutel, Alex and Smola, Alexander J. and Jing, How},
biburl = {https://www.bibsonomy.org/bibtex/2fd4e74238bf7d7b5973c05f74a7adc89/nosebrain},
booktitle = {Proceedings of the Tenth ACM International Conference on Web Search and Data Mining},
doi = {10.1145/3018661.3018689},
interhash = {6c115d03d5066bcb8a201ed19e8fab19},
intrahash = {fd4e74238bf7d7b5973c05f74a7adc89},
isbn = {978-1-4503-4675-7},
keywords = {recommendation rnn sys:toread},
location = {Cambridge, United Kingdom},
numpages = {9},
pages = {495--503},
publisher = {ACM},
series = {WSDM '17},
timestamp = {2017-04-07T17:48:13.000+0200},
title = {Recurrent Recommender Networks},
url = {http://doi.acm.org/10.1145/3018661.3018689},
year = 2017
}