Personalized Ranking for Non-Uniformly Sampled Items

Zeno Gantner, Lucas Drumond, Christoph Freudenthaler, Lars Schmidt-Thieme
Proceedings of KDD Cup 2011, PMLR 18:231-247, 2012.

Abstract

We develop an adapted version of the Bayesian Personalized Ranking (BPR) optimization criterion (Rendle et al., 2009) that takes the non-uniform sampling of negative test items – as in track 2 of the KDD Cup 2011 – into account. Furthermore, we present a modified version of the generic BPR learning algorithm that maximizes the new criterion. We use it to train ranking matrix factorization models as components of an ensemble. Additionally, we combine the ranking predictions with rating prediction models to also take into account rating data. With an ensemble of such combined models, we ranked 8th (out of more than 300 teams) in track 2 of the KDD Cup 2011, without using the additional taxonomic information offered by the competition organizers.

Cite this Paper


BibTeX
@InProceedings{pmlr-v18-gantner12a, title = {Personalized Ranking for Non-Uniformly Sampled Items}, author = {Gantner, Zeno and Drumond, Lucas and Freudenthaler, Christoph and Schmidt-Thieme, Lars}, booktitle = {Proceedings of KDD Cup 2011}, pages = {231--247}, year = {2012}, editor = {Dror, Gideon and Koren, Yehuda and Weimer, Markus}, volume = {18}, series = {Proceedings of Machine Learning Research}, month = {21 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v18/gantner12a/gantner12a.pdf}, url = {https://proceedings.mlr.press/v18/gantner12a.html}, abstract = {We develop an adapted version of the Bayesian Personalized Ranking (BPR) optimization criterion (Rendle et al., 2009) that takes the non-uniform sampling of negative test items – as in track 2 of the KDD Cup 2011 – into account. Furthermore, we present a modified version of the generic BPR learning algorithm that maximizes the new criterion. We use it to train ranking matrix factorization models as components of an ensemble. Additionally, we combine the ranking predictions with rating prediction models to also take into account rating data. With an ensemble of such combined models, we ranked 8th (out of more than 300 teams) in track 2 of the KDD Cup 2011, without using the additional taxonomic information offered by the competition organizers.} }
Endnote
%0 Conference Paper %T Personalized Ranking for Non-Uniformly Sampled Items %A Zeno Gantner %A Lucas Drumond %A Christoph Freudenthaler %A Lars Schmidt-Thieme %B Proceedings of KDD Cup 2011 %C Proceedings of Machine Learning Research %D 2012 %E Gideon Dror %E Yehuda Koren %E Markus Weimer %F pmlr-v18-gantner12a %I PMLR %P 231--247 %U https://proceedings.mlr.press/v18/gantner12a.html %V 18 %X We develop an adapted version of the Bayesian Personalized Ranking (BPR) optimization criterion (Rendle et al., 2009) that takes the non-uniform sampling of negative test items – as in track 2 of the KDD Cup 2011 – into account. Furthermore, we present a modified version of the generic BPR learning algorithm that maximizes the new criterion. We use it to train ranking matrix factorization models as components of an ensemble. Additionally, we combine the ranking predictions with rating prediction models to also take into account rating data. With an ensemble of such combined models, we ranked 8th (out of more than 300 teams) in track 2 of the KDD Cup 2011, without using the additional taxonomic information offered by the competition organizers.
RIS
TY - CPAPER TI - Personalized Ranking for Non-Uniformly Sampled Items AU - Zeno Gantner AU - Lucas Drumond AU - Christoph Freudenthaler AU - Lars Schmidt-Thieme BT - Proceedings of KDD Cup 2011 DA - 2012/06/01 ED - Gideon Dror ED - Yehuda Koren ED - Markus Weimer ID - pmlr-v18-gantner12a PB - PMLR DP - Proceedings of Machine Learning Research VL - 18 SP - 231 EP - 247 L1 - http://proceedings.mlr.press/v18/gantner12a/gantner12a.pdf UR - https://proceedings.mlr.press/v18/gantner12a.html AB - We develop an adapted version of the Bayesian Personalized Ranking (BPR) optimization criterion (Rendle et al., 2009) that takes the non-uniform sampling of negative test items – as in track 2 of the KDD Cup 2011 – into account. Furthermore, we present a modified version of the generic BPR learning algorithm that maximizes the new criterion. We use it to train ranking matrix factorization models as components of an ensemble. Additionally, we combine the ranking predictions with rating prediction models to also take into account rating data. With an ensemble of such combined models, we ranked 8th (out of more than 300 teams) in track 2 of the KDD Cup 2011, without using the additional taxonomic information offered by the competition organizers. ER -
APA
Gantner, Z., Drumond, L., Freudenthaler, C. & Schmidt-Thieme, L.. (2012). Personalized Ranking for Non-Uniformly Sampled Items. Proceedings of KDD Cup 2011, in Proceedings of Machine Learning Research 18:231-247 Available from https://proceedings.mlr.press/v18/gantner12a.html.

Related Material