Model-based learning from preference data

Qinghua Liu (1) , Marta Crispino (2) , Ida Scheel (1) , Valeria Vitelli (1) , Arnoldo Frigessi (1)

1 UiO - University of Oslo
2 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems

Preference data occurs when assessors express comparative opinions about a set of items, by rating, ranking, pair comparing, liking or clicking. The purpose of preference learning is to (i) infer on the shared consensus preference of a group of users, sometimes called rank aggregation; or (ii) estimate for each user her individual ranking of the items, when the user indicates only incomplete preferences; this is an important part of recom-mender systems. We provide an overview of probabilistic approaches to preference learning, including the Mallows, Plackett-Luce, Bradley-Terry models and collaborative filtering, and some of their variations. We illustrate , compare and discuss the use of these models by means of an experiment in which assessors rank potatoes, and with a simulation. The purpose of this paper is not to recommend the use of one best method, but to present a palette of different possibilities for different questions and different types of data.

Mots clés: Mallows model; Plackett-Luce model; Bradley-Terry model; recommender systems; Preference learning with uncertainty; Bayesian inference

Domaines: Méthodologie [stat.ME]

Qinghua Liu, Marta Crispino, Ida Scheel, Valeria Vitelli, Arnoldo Frigessi. Model-based learning from preference data. Annual Reviews of Statistics and its Application, 2019, 6 (1), pp.329-354. ⟨10.1146/annurev-statistics-031017-100213⟩. ⟨hal-01972948⟩