A Topic Modeling Approach to Ranking

Weicong Ding, Prakash Ishwar, Venkatesh Saligrama
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:214-222, 2015.

Abstract

We propose a topic modeling approach to the prediction of preferences in pairwise comparisons. We develop a new generative model for pairwise comparisons that accounts for multiple shared latent rankings that are prevalent in a population of users. This new model also captures inconsistent user behavior in a natural way. We show how the estimation of latent rankings in the new generative model can be formally reduced to the estimation of topics in a statistically equivalent topic modeling problem. We leverage recent advances in the topic modeling literature to develop an algorithm that can learn shared latent rankings with provable consistency as well as sample and computational complexity guarantees. We demonstrate that the new approach is empirically competitive with the current state-of-the-art approaches in predicting preferences on some semi-synthetic and real world datasets.

Cite this Paper

BibTeX
@InProceedings{pmlr-v38-ding15, title = {{A Topic Modeling Approach to Ranking}}, author = {Ding, Weicong and Ishwar, Prakash and Saligrama, Venkatesh}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {214--222}, year = {2015}, editor = {Lebanon, Guy and Vishwanathan, S. V. N.}, volume = {38}, series = {Proceedings of Machine Learning Research}, address = {San Diego, California, USA}, month = {09--12 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v38/ding15.pdf}, url = {https://proceedings.mlr.press/v38/ding15.html}, abstract = {We propose a topic modeling approach to the prediction of preferences in pairwise comparisons. We develop a new generative model for pairwise comparisons that accounts for multiple shared latent rankings that are prevalent in a population of users. This new model also captures inconsistent user behavior in a natural way. We show how the estimation of latent rankings in the new generative model can be formally reduced to the estimation of topics in a statistically equivalent topic modeling problem. We leverage recent advances in the topic modeling literature to develop an algorithm that can learn shared latent rankings with provable consistency as well as sample and computational complexity guarantees. We demonstrate that the new approach is empirically competitive with the current state-of-the-art approaches in predicting preferences on some semi-synthetic and real world datasets.} }
Endnote
%0 Conference Paper %T A Topic Modeling Approach to Ranking %A Weicong Ding %A Prakash Ishwar %A Venkatesh Saligrama %B Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2015 %E Guy Lebanon %E S. V. N. Vishwanathan %F pmlr-v38-ding15 %I PMLR %P 214--222 %U https://proceedings.mlr.press/v38/ding15.html %V 38 %X We propose a topic modeling approach to the prediction of preferences in pairwise comparisons. We develop a new generative model for pairwise comparisons that accounts for multiple shared latent rankings that are prevalent in a population of users. This new model also captures inconsistent user behavior in a natural way. We show how the estimation of latent rankings in the new generative model can be formally reduced to the estimation of topics in a statistically equivalent topic modeling problem. We leverage recent advances in the topic modeling literature to develop an algorithm that can learn shared latent rankings with provable consistency as well as sample and computational complexity guarantees. We demonstrate that the new approach is empirically competitive with the current state-of-the-art approaches in predicting preferences on some semi-synthetic and real world datasets.
RIS
TY - CPAPER TI - A Topic Modeling Approach to Ranking AU - Weicong Ding AU - Prakash Ishwar AU - Venkatesh Saligrama BT - Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics DA - 2015/02/21 ED - Guy Lebanon ED - S. V. N. Vishwanathan ID - pmlr-v38-ding15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 214 EP - 222 L1 - http://proceedings.mlr.press/v38/ding15.pdf UR - https://proceedings.mlr.press/v38/ding15.html AB - We propose a topic modeling approach to the prediction of preferences in pairwise comparisons. We develop a new generative model for pairwise comparisons that accounts for multiple shared latent rankings that are prevalent in a population of users. This new model also captures inconsistent user behavior in a natural way. We show how the estimation of latent rankings in the new generative model can be formally reduced to the estimation of topics in a statistically equivalent topic modeling problem. We leverage recent advances in the topic modeling literature to develop an algorithm that can learn shared latent rankings with provable consistency as well as sample and computational complexity guarantees. We demonstrate that the new approach is empirically competitive with the current state-of-the-art approaches in predicting preferences on some semi-synthetic and real world datasets. ER -
APA
Ding, W., Ishwar, P. & Saligrama, V.. (2015). A Topic Modeling Approach to Ranking. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:214-222 Available from https://proceedings.mlr.press/v38/ding15.html.

Related Material