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 global rankings that are prevalent in a population of users. This new model also captures inconsistent user behavior in a natural way. We establish a formal statistical equivalence between the new generative model and topic models. 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.
This is joint work with Weicong Ding and Prakash Ishwar.