Permutation Models for Collaborative Ranking

TL;DR

This paper introduces new permutation-based models for collaborative ranking, extending the Plackett-Luce model with parameter factoring and community modeling, along with a log-linear approach with MCMC inference.

Contribution

It proposes novel permutation models for collaborative ranking, including extensions of Plackett-Luce and a log-linear model with efficient inference methods.

Findings

Extended Plackett-Luce with user-specific parameters

Modeled community influence in ranking predictions

Developed MCMC-based learning and linear-time prediction algorithms

Abstract

We study the problem of collaborative filtering where ranking information is available. Focusing on the core of the collaborative ranking process, the user and their community, we propose new models for representation of the underlying permutations and prediction of ranks. The first approach is based on the assumption that the user makes successive choice of items in a stage-wise manner. In particular, we extend the Plackett-Luce model in two ways - introducing parameter factoring to account for user-specific contribution, and modelling the latent community in a generative setting. The second approach relies on log-linear parameterisation, which relaxes the discrete-choice assumption, but makes learning and inference much more involved. We propose MCMC-based learning and inference methods and derive linear-time prediction algorithms.