Constructive Preference Elicitation for Multiple Users with Setwise Max-margin

Abstract

In this paper we propose an approach to preference elicitation that is suitable to large configuration spaces beyond the reach of existing state-of-theart approaches. Our setwise max-margin method can be viewed as a generalization of max-margin learning to sets, and can produce a set of "diverse" items that can be used to ask informative queries to the user. Moreover, the approach can encourage sparsity in the parameter space, in order to favor the assessment of utility towards combinations of weights that concentrate on just few features. We present a mixed integer linear programming formulation and show how our approach compares favourably with Bayesian preference elicitation alternatives and easily scales to realistic datasets.

Interactive Elicitation of a Majority Rule Sorting Model with Maximum Margin Optimization

“…Finally, another important direction is to elicit the preferences of several users, providing methods that can exploit the similarity between users, as in Teso et al [26]. Fig.…”

“…Our work is stimulated by recently proposed approaches for eliciting multi-attribute utility functions using maximum-margin optimization [25,26] in configuration problems. The maximum-margin optimization that we adopt has been used (with some variations) in previous works [21,22] that tackled the problem of learning the weights of a MR-sort model; these works, however, did not consider incremental elicitation.…”

Constructive Preference Elicitation

“…Our evaluation shows that SM, PP, and CPP can easily solve problems much larger than previously possible, while still performing as well or better than state-of-the-art competitors. Please note that, while SM and CPP have previously been presented in Teso et al (2016) and Teso et al (2017a), the application of PP to layout synthesis is a novel contribution of this article.…”

“…SM (or SM) (Teso et al, 2016) is an implementation of our CPE framework which generalizes the max-margin principle to sets to recommend both high-quality and maximally diverse configurations to the DM.…”

“…We showcase the feasibility of our framework by surveying three concrete implementations. The first one is a setwise generalization of max-margin learning (SetMargin or SM for short) (Teso et al, 2016). It relies on setwise ranking queries, an extension of pairwise ranking queries to small collections of alternatives, for increased flexibility.…”