Social Choice Random Utility Models of Intransitive Pairwise Comparisons
This addresses the need for more realistic models of human choice behavior in applications like recommendation systems and elections, though it is incremental as it builds on existing parametric and social choice frameworks.
The authors tackled the problem of modeling intransitive pairwise preferences in discrete choice models, showing that no parametric model can have a concave log-likelihood while capturing intransitivity, and proposed a Majority Vote model based on social choice theory that uses multidimensional embeddings to represent arbitrary intransitive cycles and pairwise probabilities.
There is a growing need for discrete choice models that account for the complex nature of human choices, escaping traditional behavioral assumptions such as the transitivity of pairwise preferences. Recently, several parametric models of intransitive comparisons have been proposed, but in all cases the maximum likelihood problem is non-concave, making inference difficult. In this work we generalize this trend, showing that there cannot exist any parametric model with a concave log-likelihood function that can exhibit intransitive preferences. Given this result, we motivate a new model for analyzing intransitivity in pairwise comparisons, taking inspiration from the Condorcet method (majority vote) in social choice theory. The Majority Vote model we analyze is defined as a voting process over independent Random Utility Models (RUMs). We infer a multidimensional embedding of each object or player, in contrast to the traditional one-dimensional embedding used by models such as the Thurstone or Bradley-Terry-Luce (BTL) models. We show that a three-dimensional majority vote model is capable of modeling arbitrarily strong and long intransitive cycles, and can also represent arbitrary pairwise comparison probabilities on any triplet. We provide experimental results that substantiate our claims regarding the effectiveness of our model in capturing intransitivity for various pairwise choice tasks such as predicting choices in recommendation systems, winners in online video games, and elections.